Estimate of ischemic stroke prevalence according to a novel 4-tiered classification of left ventricular hypertrophy: insights from the general Chinese population

Abstract

Background: Recently, a novel 4-tiered classification of left ventricular hypertrophy (LVH) based on ventricular dilatation (indexed LV end-diastolic volume [EDV]) and concentricity (mass/EDV^0.67) has improved all-cause and cardiovascular mortality risk stratification. However, their possible association with ischemic stroke has not been extensively evaluated in the general population.

Methods: We evaluated a cross-sectional study of 11,037 subjects from the general population of China in whom echocardiographic and ischemic stroke data were available to subdivide patients with LVH into four geometric patterns: indeterminate, dilated, thick and both thick and dilated hypertrophy.

Results: Compared with normal LV geometry, indeterminate and thick hypertrophy showed a higher prevalence of ischemic stroke (p < .05). Ischemic stroke was significantly greater in participants with indeterminate (adjusted odd ratio [OR]:1.635, 95% confidence interval [CI]: 1.115–2.398) and thick (2.143 [1.329–3.456]) hypertrophy but not significantly in those with dilated (1.251 [0.803–1.950]) and both thick and dilated hypertrophy (0.926 [0.435–1.971]) compared with normal geometry in multivariable analysis.

Conclusions: Indeterminate and thick hypertrophy were significantly associated with the presence of ischemic stroke in the general Chinese population. The new 4-tiered categorization of LVH can permit a better understanding of which subjects are at high enough risk for ischemic stroke to warrant early targeted therapy.

Key messages

• This was the first study to investigate whether a 4-tiered classification of LVH defines subgroups in the general population that are at variable risks of ischemic stroke.

• We identified that thick hypertrophy carried the greatest odd for ischemic stroke, independently of traditional risk factors, followed by indeterminate hypertrophy.

• The new 4-tiered categorization of LVH emerged as a valuable operational approach, a potential alternative to LVM, to refine ischemic stroke stratification in general population.

Keywords:

• Hypertrophy
• left ventricular geometry
• ischemic stroke
• echocardiography
• epidemiology
• general population

1. Introduction

Left ventricular hypertrophy (LVH), a target organ response to hypertension, is considered an important indicator of subclinical cardiovascular disease (CVD) and is a manifestation of the pathophysiologic basis for cardiac remodeling strongly correlated with adverse cardiovascular events [1–3]. Traditional two-tiered classification of LVH can be distinguished based on whether left ventricular mass (LVM) and relative wall thickness (a ratio of LV wall thickness to LV chamber dimension) are normal versus increased, comprising three pathological LV patterns, namely concentric remodeling, eccentric hypertrophy and concentric hypertrophy [4]. Though widely used, the major limitations of this paradigm have been the lack of accounting for LV dilation in isolation, a critical aspect of geometric remodeling, and the absence of a specific category identifying individuals with concurrent LVH and dilated LV chamber. The capability of this classification system to confer cardiovascular risk and provide prognostic information beyond LVM has been conflicting [3,5,6].

Recently, the new 4-tiered classification system for LVH based on LV concentricity^0.67 [LVM/LV end-diastolic volume (LVEDV)^0.67] and LV dilation (LVEDV/body surface area [BSA]) predisposed individuals at differential risk of unfavorable cardiovascular outcomes [7–12]. In this intriguing classification, subjects with eccentric hypertrophy were stratified into indeterminate hypertrophy and dilated hypertrophy. Similarly, concentric hypertrophy could exist in nondilated or dilated forms, termed as thick hypertrophy and both thick and dilated hypertrophy. This approach provided a more refined assessment of the geometric patterns of increased LVM and enhanced the prediction of worse prognosis from the standard 2-group categorization of LVH [13]. For instance, the investigators of the Dallas Heart Study revealed that the cumulative incidence of heart failure or cardiovascular death was increased in those with dilated, thick, or both thick and dilated hypertrophy, but not in indeterminate hypertrophy in comparison with those without LVH among the general population [8]. Bang et al. tested the prognostic value of such new paradigm in 939 hypertensive patients, yielding consistent results with respect to its ability to predict all-cause mortality or cardiovascular events [9].

Stroke is a major global health problem and currently the second largest contributor to death and the third-leading cause of disability [14]. Ischemic stroke is the most common subtype worldwide and has a substantial case fatality rate [14,15]. Broadly, transthoracic echocardiography has been recognized to reflect the presence of cardioembolic sources in the setting of diagnosing ischemic stroke patients [16]. There is a wealth of evidence showing that LVH represents one of the main risk factors responsible for stroke [17–20]. Specifically, one study reported that concentric LVH heralded the highest risk of developing ischemic stroke in cases of abnormal LV geometry [21]. In spite of these prior researches, there are no data on the predictive value of novel 4-group categorization of LVH for the identification of ischemic stroke. Given that improved subphenotypic characterization identified patterns of LVH that appeared to represent differences in specific LV mechanics and cardiovascular prognosis [7–12,22], it can provide important clues in efforts to determine whether a thick-walled LV has concomitant LV dilatation or not will enhance the prediction of ischemic stroke risk. Hence, our aim in this study was to examine the association between 4-tiered classification system for LVH, detected by echocardiography, and the presence of ischemic stroke among 11,037 participants of the general population from China.

2. Methods

2.1. Study population

This study was derived from a population-based cross-sectional epidemiological investigation evaluating the presence of cardiovascular risk factors in 11,956 permanent residents (≥ 35 years of age) of northeast China. Details regarding the rationale, design, and implementation have been described previously [23–25]. Our eligible datasets (echocardiographic and clinical data) consisted of 11,037 participants in the current analysis. The Ethics Committee of China Medical University (Shenyang, China) approved the study protocol. Subjects who passed the initial screening examination were enrolled in our study after providing written informed consent and the whole data and procedures conformed to the Declaration of Helsinki and its subsequent modifications.

2.2. Data collection and measurements

Data collection and methods were employed as described elsewhere [23–25]. For all participants, measurements were performed and data were collected at the time of study entry. Cardiologists and trained nurses conducted face-to-face interviews using a structured questionnaire to document specified data on sociodemographic characteristics, physical activity, cigarette smoking, alcohol consumption, and history of CVD (coronary heart disease, atrial fibrillation, other arrhythmia, heart failure, and stroke). All patients underwent the following procedures: resting blood pressure measurements, fasting blood draw, and anthropometric assessments. Individuals with ischemic stroke and ischemic stroke-free subjects underwent two-dimensional transthoracic echocardiographic examination

Study participants waited for at least 5 min in a relaxed and sitting position. Then blood pressure (BP) was measured by a standardized automatic electronic sphygmomanometer (HEM-907; Omron, Kyoto, Japan) using an appropriately sized cuff with the arm supported at the level of the heart. The mean readings of three replicate measurements were recorded for the present analysis. Body mass index (BMI) was quantified as body weight divided by the square of height (kg/m²). The subjects were reminded of keeping an overnight fasting with 12 h before the investigation. Venous blood specimens were designed to be obtained for assessing total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and fasting plasma glucose (FPG). A full description of comprehensive storage process and standard laboratory measurement methods have been published in prior reports [23–25].

Based on strict neurological examination, computed tomography or magnetic resonance imaging, first-ever ischemic stroke was defined as stroke event with a diagnosis of thrombosis or embolism. Documentation of ischemic stroke reviewed by two independent neurologists was confirmed on the basis of the information on diagnostic tests or hospital records. The diagnosis of hypertension was established as BP level of at least 140/90 mmHg, individuals who were on antihypertensive medications or a prior diagnosis of hypertension. Diabetes was defined as the patient’s self-reported history of diabetes, use of diabetes medications or an FPG ≥7 mmol/L.

2.3. Echocardiographic analysis

The protocol for performing and reading transthoracic echocardiograms has been previously described [26] and are in accordance with recommendations from the American Society of Echocardiography [27]. LVM was calculated based on the Devereux formula [28]: LVM = 0.80 × (1.04 × [(IVSTd + LVIDd + PWTd)³ – LVIDd³]) + 0.6, where IVSTd is interventricular septum thickness at end diastole, LVIDd is LV end-diastolic internal diameter, and PWTd is posterior wall thickness at end diastole. LV chamber volumes and ejection fraction by the angiographically validated Teichholz method. For the primary analysis, LVH and increased LVEDV/BSA were defined according to guidelines (LVM/BSA: ≥ 96 g/m² [females] and ≥116 g/m² [males]; EDV/BSA: ≥76 ml/m² [9,29,30]. Previously defined thresholds for LV concentricity^0.67 (≥ 8.1 g/ml^0.67 for females and ≥9.1 g/ml^0.67 for males) were used [7–9]. This approach yields new 4 categories of LVH: (1) indeterminate hypertrophy (neither increased concentricity^0.67 nor increased LVEDV/BSA); (2) dilated hypertrophy (increased LVEDV/BSA without increased concentricity^0.67); (3) thick hypertrophy (increased concentricity^0.67 without increased LVEDV/BSA); and (4) both thick and dilated hypertrophy (increased concentricity^0.67 and LVEDV/BSA). A sensitivity analysis was performed by indexing LVM to height^2.7 using thresholds of ≥39 g/m^2.7 for females and ≥48 g/m^2.7 for males [7,8].

2.4. Statistical analysis

Descriptive data are reported as mean ± SD, and frequencies are percentages. Continuous variables without normal distribution were expressed as median with first and third quartiles. For comparison between groups, the chi-square test or Fisher exact test was used for categorical variables and unpaired Student t test for continuous variables, as appropriate. Differences for continuous variables among more than 2 groups were evaluated using one-way analysis of variance, Kruskal-Wallis and Bonferroni post hoc test. To test whether the 4-tiered classification system was clinical relevant, logistic regression analysis was used to estimate odds ratio (OR) and 95% confidence intervals (95% CI) for the risk of having ischemic stroke in the presence of abnormal LV geometric patterns, with subjects with normal LVM as the reference group. The association between updated classification of LVH with ischemic stroke were examined in an unadjusted fashion (model 1), after adjustment for age, sex, race, physical activity, current smoking and drinking status (model 2), and after additional adjustment for BMI, TC, hypertension, diabetes, and LVM/BSA (model 3). Three sensitivity analyses were constructed, including repeating the primary analyses using LVH defined by LVM indexed to height^2.7; using the continuous parameter of systolic blood pressure (SBP) instead of hypertension as a covariate; and including subjects with a history of CVD. All statistical tests were two-sided, and results with p values < .05 were considered significant. SPSS 22.0 software (IBM Corp) was applied for data analysis.

3. Results

The age of our 11,037 participants was 53.7 ± 10.5 years, with 54.0% females. On the basis of the new 4-tiered classification system of LVH, 1254 subjects (11.3%) had LVH, of whom 484 (4.4%) were classified as indeterminate hypertrophy, 386 (3.5%) as dilated hypertrophy, 246 (2.2%) as thick hypertrophy and 138 (1.3%) as having both thick and dilated hypertrophy (Table 1). Differences in age were observed among the abnormal LV geometric patterns, especially when associated with LV dilation. Indeterminate hypertrophy group had significantly more females. There were higher BP in the concentric groups without difference in HDL-C levels and prevalence of current smokers and diabetes. Moreover, the novel categorization system captured a higher LDL-C in the thick hypertrophy group; less frequently taking alcohol among those with concentric LVH in comparison with indeterminate group; and higher diastolic blood pressure, BMI, TC, TG, and prevalence of hypertension in both concentric groups compared with dilated hypertrophy group.

Table 1. Characteristics stratified by the presence of LVH and four-tiered geometric patterns of LVH.

		LVH
		Eccentric (n = 860)		Concentric (n = 384)
Variables	No LVH (n = 9783)	Indeterminate (n = 484)	Dilated (n = 386)	Thick (n = 246)	Both (n = 138)	p value*
Age (years)	53.11 ± 10.34	59.08 ± 9.57^a	61.19 ± 10.29^a,b	58.06 ± 10.56^a,c	61.03 ± 10.76^a	< .001
Female, n (%)	5148 (52.6)	396 (81.8)^a	197 (51.0)^b	147 (59.8)^b	70 (50.7)^b	< .001
Race (Han), n (%)	9276 (94.8)	462 (95.5)	377 (97.7)	231 (93.9)	132 (95.7)	.124
Physical activity, n (%)						.166
Low	8562 (87.5)	416 (86.0)	330 (85.5)	213 (86.6)	113 (81.9)
Moderate	848 (8.7)	46 (9.5)	46 (11.9)	22 (8.9)	16 (11.6)
High	373 (3.8)	22 (4.5)	10 (2.6)	11 (4.5)	9 (6.5)
Current smoker, n (%)	3483 (35.6)	134 (27.7)^a	139 (36.0)	79 (32.1)	47 (34.1)	.008
Current drinker, n (%)	2265 (23.2)	58 (12.0)^a	71 (18.4)	52 (21.1)^b	30 (21.7)^b	< .001
SBP (mmHg)	139.19 ± 21.76	158.42 ± 26.32^a	155.79 ± 24.56^a	168.06 ± 29.27^a,b,c	166.72 ± 24.37^a,b,c	< .001
DBP (mmHg)	81.22 ± 11.16	87.07 ± 13.70^a	85.58 ± 12.82^a	93.06 ± 15.32^a,b,c	90.13 ± 15.68^a,c	< .001
FPG (mmol/L)	5.88 ± 1.60	6.21 ± 1.90^a	6.01 ± 1.76	6.30 ± 2.08^a	6.31 ± 1.71^a	< .001
BMI (kg/m^2)	24.67 ± 3.59	25.70 ± 3.95^a	24.82 ± 4.05^b	26.83 ± 4.12^a,b,c	26.40 ± 3.71^a,c	< .001
TC (mmol/L)	5.21 ± 1.07	5.53 ± 1.08^a	5.26 ± 1.07^b	5.55 ± 1.33^a,c	5.63 ± 1.34^a,c	< .001
TG (mmol/L)	1.22 (0.87–1.86)	1.52 (1.06–2.26)^a	1.27 (0.91–1.94)^b	1.47 (1.06–2.27)^a,c	1.52 (1.09–2.25)^a,c	< .001
LDL-C (mmol/L)	2.91 ± 0.81	3.16 ± 0.83^a	2.91 ± 0.81^b	3.22 ± 0.97^a,c	3.06 ± 0.88	< .001
HDL-C (mmol/L)	1.41 ± 0.38	1.39 ± 0.37	1.38 ± 0.36	1.35 ± 0.34	1.34 ± 0.31	.007
Hypertension, n (%)	4582 (46.8)	395 (81.6)^a	296 (76.7)^a	214 (87.0)^a,c	127 (92.0)^a,b,c	< .001
Diabetes, n (%)	941 (9.6)	92 (19.0)^a	48 (12.4)	48 (19.5)^a	29 (21.0)^a	< .001
History of CVD, n (%)	1521 (15.5)	130 (26.9)^a	116 (30.1)^a	86 (35.0)^a	47 (34.1)^a	< .001
Prevalence of ischemic stroke, n (%)	251 (2.6)	36 (7.4)^a	25 (6.5)^a	25 (10.2)^a	8 (5.8)	< .001
Echocardiographic parameters
LV wall thickness (mm)	8.46 ± 0.88	9.92 ± 0.69^a	9.57 ± 0.88^a	16.26 ± 10.71^a,b,c	12.79 ± 5.75^a,b,c,d	< .001
Relative wall thickness	0.36 (0.33–0.38)	0.41 (0.38–0.43)^a	0.35 (0.32–0.38)^a,b	0.48 (0.44–0.53)^a,b,c	0.40 (0.38–0.43)^a,c,d	< .001
LVM/BSA (g/m^2)	77.48 (68.77–86.94)	103.48 (98.87–113.97)^a	120.82 (109.71–130.30)^a,b	127.99 (117.13–145.96)^a,b	152.62 (141.03–165.38)^a,b	< .001
LVM/height^2.7 (g/m^2.7)	35.74 (31.34–40.78)	51.83 (47.98–55.51)^a	57.13 (52.45–62.78)^a	61.00 (55.04–71.59)^a	72.66 (66.08–82.26)^a,b	< .001
Concentricity^0.67 (g/ml^0.67)	5.69 (5.56–6.54)	7.52 (6.97–8.04)^a	7.14 (6.58–7.70)^a	9.69 (9.12–11.48)^a,b,c	9.22 (8.66–9.80)^a,b,c	< .001
LVEDV (ml)	101.29 ± 20.25	110.38 ± 14.76^a	144.81 ± 39.20^a,b	102.75 ± 27.68^b,c	155.13 ± 44.68^a,b,c,d	< .001
LVEDV/BSA (ml/m^2)	60.55 ± 10.24	68.57 ± 5.19^a	90.04 ± 24.11^a,b	59.30 ± 14.12^b,c	91.11 ± 25.00^a,b,d	< .001
Stroke volume (ml)	63.11 ± 16.26	67.60 ± 14.36^a	89.08 ± 34.91^a,b	64.02 ± 18.33^c	99.30 ± 42.80^a,b,c,d	< .001
LVEF (%)	62.06 ± 9.98	61.30 ± 10.23	61.83 ± 12.44	60.28 ± 13.59	63.35 ± 8.72	.019

Data are expressed as mean ± standard deviation or median (interquartile range) for continuous variables, and frequencies (percentages) for categorical variables.

SBP: indicates systolic blood pressure; DBP: diastolic blood pressure; BMI: body mass index; TC: total cholesterol; TG: triglyceride; LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; LVM: left ventricular mass; BSA: body surface area; LVEDV: left ventricular end diastolic volume; LVEF: left ventricular ejection fraction; LVH: left ventricular hypertrophy.; *For comparison among the four-tiered geometric patterns of LVH.

^ap < .05 versus no LVH group.

^bp < .05 versus indeterminate hypertrophy group.

^cp < .05 versus dilated hypertrophy group.

^dp < .05 versus thick hypertrophy group.

In comparing subjects with thick hypertrophy with other subjects with LVH, the former had a significantly higher LV wall thickness and RWT (Table 1). LVM index (LVMI) was increased in all abnormal LV geometry patterns, but it progressively increased from normal LV geometry, across the subjects with indeterminate, dilated, and thick hypertrophy, to both thick and dilated hypertrophy. The same pattern was shown in LVEDV/BSA and stroke volume, which were consistently higher in subjects with LV dilation associated with either eccentric or concentric LVH. Collectively, participants with concentric LVH versus eccentric LVH were more likely to have an increased concentricity^0.67.

The subclassification of subjects with eccentric LVH into groups with normal (7.4%) or elevated LV chamber volume (6.5%) revealed that both indeterminate and dilated hypertrophy had a higher prevalence of ischemic stroke as compared with those without LVH (2.6%) (Figure 1). In contrast, when concentric LVH was subdivided into two subgroups (thick hypertrophy or both thick and dilated hypertrophy), the former, but not the latter, exhibited significantly higher frequencies of ischemic stroke than normal LV geometry (10.2% vs 2.6%).

Figure 1. Prevalence of ischemic stroke stratified by 2- and 4-tired classification of LVH. *p < .05 versus no LVH group. LVH indicates left ventricular hypertrophy.

In multivariable analysis, using the 2-tiered model, both eccentric and concentric LVH patients experienced 44% and 59% higher odds, respectively, of having ischemic stroke (Table 2). When applying the 4-group classification of LVH, for subjects with indeterminate hypertrophy, odds of ischemic stroke were more than half for those without LVH (OR, 1.635; 95% CI, 1.115–2.398; p = .012), whereas those with dilated hypertrophy were not significantly associated with ischemic stroke. Likewise, among patients with concentric LVH, thick hypertrophy entailed a two-fold higher odd of having ischemic stroke than was seen in those with a normal LVM (OR, 2.143; 95% CI, 1.329–3.456; p = .002), but the both thick and dilated hypertrophy group did not. Additionally, the continuous parameters of LV concentricity^0.67 (OR, 1.067; 95% CI, 1.024–1.113 per 1 SD increment) was independently associated with the presence of ischemic stroke in multivariable analysis adjusted for age, sex, race, physical activity, current smoking and drinking status, BMI, TC, hypertension and diabetes, while LVEDV/BSA was not (OR, 0.957; 95% CI, 0.859–1.065 per 1 SD increment).

Table 2. Unadjusted and multivariable adjusted associations of 2- and 4-tired classification of LVH with ischemic stroke.

	Model 1		Model 2		Model 3
LVH classification	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value
Two-tiered
No LVH*	1.000 (reference)		1.000 (reference)		1.000 (reference)
Eccentric LVH	2.863 (2.145–3.822)	< .001	1.804 (1.332–2.442)	< .001	1.444 (1.066–1.956)^a	.018
Concentric LVH	3.570 (2.446–5.212)	< .001	2.315 (1.560–3.435)	< .001	1.594 (1.070–2.374)^a	.022
Four-tired
No LVH*	1.000 (reference)		1.000 (reference)		1.000 (reference)
Indeterminate	3.052 (2.125–4.382)	< .001	2.151 (1.471–3.145)	< .001	1.635 (1.115–2.398)	.012
Dilated	2.630 (1.721–4.020)	< .001	1.454 (0.936–2.259)	.096	1.251 (0.803–1.950)	.323
Thick	4.296 (2.789–6.618)	< .001	3.120 (1.986–4.902)	< .001	2.143 (1.329–3.456)	.002
Both thick and dilated	2.337 (1.132–4.825)	.022	1.255 (0.595–2.648)	.551	0.926 (0.435–1.971)	.841

LVH is indexed BSA.

*Referent group. Model 1: unadjusted.

Model 2: adjusted for age, sex, race, physical activity, current smoking, and drinking status.

Model 3: adjusted for all the factors in Model 2 and body mass index, total cholesterol, hypertension, diabetes, and LVM/BSA. OR indicates odd ratio.

CI: confidence interval; LVH: left ventricular hypertrophy.

^aThe model did not include LVM/BSA in the multivariate analysis of two-tiered LVH and ischemic stroke.

In the first sensitivity analyses in which LVH was defined based on LVM indexed to height^2.7 (Supplementary Table 1), there were 7915 (71.7%) participants without LVH, 2129 (19.3%) with indeterminate hypertrophy, 585 (5.3%) with dilated hypertrophy, 270 (2.4%) with thick hypertrophy, and 138 (1.3%) with both thick and dilated hypertrophy with this approach. The majority of key findings depicted above persisted when using LVM/height^2.7 to define LVH. Specifically, in the group with eccentric LVH, prevalent ischemic stroke was more common in those with indeterminate hypertrophy than those without LVH (5.0% vs 2.2%, respectively, OR, 1.383; 95% CI, 1.034–1.850; p = .029). Importantly, among those formerly classified as having concentric LVH, only thick hypertrophy remained correlated with ischemic stroke in the fully-adjusted model (OR, 2.237; 95% CI, 1.387–3.607; p = .001). Additional sensitivity analyses including subjects with a history of CVD and using SBP as a covariate in place of hypertension revealed similar associations (Supplementary Tables 2 and 3).

4. Discussion

In a large general population from China, incorporation of LV dilatation into the assessment of LVH identified important subphenotypes within the standard two-tiered categorization that have differential risk. In essence, the presence of thick hypertrophy and indeterminate hypertrophy entailed greater odds of ischemic stroke. Unexpectedly, dilated hypertrophy and both thick and dilated hypertrophy were not associated with increased prevalence of ischemic stroke. These findings extended previous reports and provided a rationale for the observed unfavorable prognosis in patients with indeterminate and thick hypertrophy, further supporting the role of new geometric forms of LVH in the assessment of cardio-cerebrovascular risk. Enhanced LVH characterization that accounted for LV wall thickness in addition to LV dilatation would offer the potential to apply strategies directed at indeterminate and thick hypertrophy, efforts which may prevent incident ischemic stroke and improve cardiovascular outcomes.

LV remodeling is generally thought to be the process of structural changes in the LV in response to changes in intrinsic myocardial tissue characteristics and architecture, or to extraneous stimuli caused by increased pressure or volume overload [31]. Dilatation of the LV chamber or thickening of the LV walls linked the type of hemodynamic stress to increased LVM, remodeling, and hypertrophy. Abnormal LV geometry as characterized by maladaptive remodeling response to high blood pressure is considered a valuable phenotyping tool to differentiate CVD processes and has been long regarded as a pivotal phenotype in the progression of heart failure and CVD [1–3]. Prior work has largely been restricted to studying LVM and RWT as markers of LV remodeling, subdividing the population into those with normal geometry, concentric remodeling, eccentric hypertrophy, and concentric hypertrophy [27]. Despite the conventional wisdom that the paradigm proposed by Ganau has proven to be robust and conceptually appealing as evidenced by its wide use in event prediction, risk stratification and as endpoints in clinical trials [4], there is extensive evidence pointing that standard 2-group classification of LVH does not account independently for changes in LV volume and LV wall thickening, and the geometric assumptions on LVH needed to calculate volume from linear dimensions. One limitation of this classification is the lack of a specific category identifying patients with concurrent LV hypertrophy and dilated LV chamber. It should also be noted that there is consistency in the literature linking concentric LVH with poor outcomes [3,32], whereas considerable evidence has raised uncertainties regarding the adverse impact of eccentric LVH on prognosis [1,33,34].

In considering this issue, a refined 4-tiered model based on ventricular dilation and increased LV concentricity have been recently developed [7]. The LIFE study demonstrated that reclassification of hypertensive patients with LVH into four groups identified important differences in hemodynamic and LV function measures more clearly than the traditional division [30]. In keeping with this hypothesis, the value of this newly updated criteria of LV in the prediction of incident heart failure, cardiovascular events, and all-cause mortality has been uncovered, reinforcing the long-standing hope that tailoring the treatment of hypertension to cardiac phenotype, in this case LVH and its geometric subtypes determined by echocardiography or CMR, could improve outcomes in those who are at high risk [8–13]. Nevertheless, studies assessing the efficacy of this new 4-tiered classification of LVH for predicting ischemic stroke are lacking. To our knowledge, this is the first study to address this relationship in a large general population from China.

In the present investigation, indeterminate and thick hypertrophy imparted a better estimate of the presence of ischemic stroke. Such information reflected that identification of these distinct phenotypes might convey added information and clinical utility compared to standard classification. Interestingly, inconsistent with the previously mentioned reports, the novel 4-tiered categorization system of LVH failed to detect associations of dilated and both thick and dilated hypertrophy with prevalent ischemic stroke, which was certainly intriguing and suggested the need to further investigate the prognosis of patients with hypertrophy and chamber dilation to shed light on the cardiovascular complications in the general population. One explanation for these differences could lie in the characteristics of population. Prior findings were mostly limited to selected samples of hypertensive or coronary artery disease patients [9–11,22,30], whereas our current study was drawn from the general population with a relatively low cardiovascular risk. In addition, it may be partially attributed to worse cardiometabolic risk factors, such as increased levels of BP, FPG, BMI, and atherogenic lipids, for the individuals with indeterminate and thick hypertrophy compared with dilated and both thick and dilated hypertrophy, respectively. It is likely that high levels of heterogeneity for these groups may provide possible answers about different odds of ischemic stroke. Furthermore, the differing results observed in this may be related the low prevalence of abnormal LV geometry. We found that the proportion of subjects without LVH was markedly higher (88.7%) in our study than in previous reports that ranged from 24.8% to 66% in hypertensive patients [9–11,22,30] and 68.1% to 76.2% in a general population [7,8,12]. Finally, current evidence might be mixed due to important differences between Westerns and Asians in CVD disease risk factors. For example, prior researches conducted in the general population have described overall high prevalence of obesity in Whites and Blacks [7,8,12] than our sample. Because hypertension and obesity have been shown to strongly affect cardiac geometry [35], differences in their prevalence may be the cause for the differences in results from studies that have assessed this domain. However, our results, indeed, showing a significant association between indeterminate hypertrophy and ischemic stroke corroborated the PAMELA study that found a higher risk of all-cause mortality among individuals with indeterminate hypertrophy [12]. Taken together, refined subphenotypic characterization identified patterns of LVH that appeared to represent different prevalent ischemic stroke, confirming the necessity of developing a scheme that classifies specific patterns of LV structural remodeling seen in a wide spectrum of diseases states. Also of note is that the influence of thick hypertrophy on ischemic stroke was more prominent than in patients with indeterminate hypertrophy. One potential explanation is that cardiovascular resistance and cerebral perfusion were disparate in view of LV geometric pattern where patients with eccentric hypertrophy had the lowest cardiovascular resistance and highest cardiac index, which led to relatively increased cerebral perfusion [36]. Alternatively, when analyzed as continuous variables, LV concentricity but not LVEDV was a significant determinant of ischemic stroke in multivariable analysis, thus providing new insight into use of four “distinct geometric patterns” of LVH for ischemic stroke stratification in a general population. In this regard, consideration of LVM, volume, and mass/volume allows classification of LVH that includes virtually all LV remodeling changes that are seen in certain diseases. This framework paved the way for at least start deliberating that some patterns of LV architectural changes portended higher prevalence of ischemic stroke whereas others appear to be adaptive and physiologic without worsening consequences.

The pathogenic mechanisms underlying alterations in LV geometry that are associated with ischemic stroke are not entirely clear, but we hypothesize several plausible explanations. First, owing to the correlation of LVM with left atrial enlargement, subjects with LVH may be predisposed to ischemic stroke mediated by left atrial enlargement that in turn may concur to promote the formation of a clot in the atrial chamber, thereby bringing about the development of thromboembolism [37]. Atrial arrhythmias such as atrial fibrillation attributable to left atrial enlargement could have a contributory role in the occurrence of thromboembolism, an effect that is possibly lead to ischemic stroke [17,38]. Second, it is postulated that both the heart and brain are potential end organs at risk for injury driven by longstanding elevated BP indicative of arterial hypertension, which emerges as key correlates of connecting LVH and ischemic stroke [17,39,40]. Furthermore, increased LVMI tended to be the hallmark of carotid artery plaque and intima-media thickening [41,42]. Indeed, carotid disease has been assumed to parallel LVM after adjusting for conventionally measured BP and represent a particularly sensitive marker of ischemic stroke [43]. As evident, LVH is expression of the severity of aortic atheroma, therefore, implying some degree of developing vascular events [44,45].

This study should be read in the context of certain limitations. Due to the cross-sectional nature of our study, we are unable to assign causality to our findings. Further studies are warranted to evaluate the practicability and efficacy of therapy targeted to new 4-tiered cardiac structural phenotype in the prevention of ischemic stroke. Second, we did not consider the decisive role of concentric remodeling in the framework of assessing the presence of ischemic stroke given that it is not captured in the updated 4-group categorization, although conferring an adverse outcome than those with normal LVM [46]. Third, our study consisted of predominantly the general population in a particular province in China, and accordingly, the generalizability of conclusions is limited. Fourth, the subjects with ischemic stroke we observed were nonfatal stroke, not including fatal stroke, may be a selection bias. Finally, lack of information on large artery atherosclerosis, small artery occlusion, and cardiometabolic stroke restricted our ability to accurately address issues relating to 4-tiered classification of LVH as a determinant of subtypes of ischemic stroke. Nonetheless, a merit of this study was the substantial number of individuals with identifiable medical records including the clinical characteristics and echocardiographic parameters. The message of the study highlighted that it was clinically important to locate an ischemic stroke in patients with indeterminate hypertrophy and thick hypertrophy and underline the role of new classification of LV geometry in the assessment of global cardiovascular risk status. In light of this, the data presented here further supported the usefulness of routinely searching for LV nondilated hypertrophy so as to identify patients at higher risk who are supposed to undergo a more intensive pharmacological and nonpharmacological intervention to prevent the development of ischemic stroke.

5. Conclusions

In comparison with the traditional 2-tiered classification system of LVH, the novel 4-tiered categorization scheme revealed the significant associations of indeterminate and thick hypertrophy with presence of ischemic stroke in the general population, providing strong evidence of a unique, independent role of indeterminate and thick hypertrophy in cardio-cerebrovascular disease. In a practical perspective, these data raise the possibility that patients without LV chamber dilation associated with both eccentric and concentric patterns of LVH are candidates for a more refined cardiovascular risk stratification. Refinement of the phenotypic characterization of LVH may improve our understanding of its natural history and provide an opportunity for more specific and possibly earlier therapeutic intervention of ischemic stroke.

Acknowledgements

We would like to express our sincere thanks to all the authors. The authors would like to thank Dr. Yingxian Sun who was responsible for the project completion.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by grants from the “Twelfth Five-Year” project funds (National Science and Technology Support Program of China, Grant #2012BAJ18B02) and “Thirteenth Five-Year” program funds (The National Key Research and Development Program of China, Grant #2017YFC1307600).