Left ventricular mechanical dispersion in a middle-aged population: associations with subclinical cardiac remodelling and cardiovascular risk

Abstract

Background

Heterogeneous myocardial contraction is common in myocardial disease and linked to ventricular arrhythmias, but its frequency and relevance in the general population remain unclear. This study aimed to investigate the frequency of increased left ventricular mechanical dispersion (LVMD) and to examine its clinical and imaging correlates in a population-based cohort.

Methods

LVMD was assessed using speckle-tracking echocardiography in 3997 participants aged 50–64 years (51% women) from the Swedish Cardiopulmonary Bioimage Study. The upper limit of normal (ULN) for LVMD was defined as the 95th percentile in a subgroup (n=1165) without cardiovascular disease or risk factors. In the full cohort, cardiovascular risk factors, including coronary artery calcium score (CACS) and left ventricular mass index (LVMI) from cardiac CT, were compared between subjects with and without high LVMD, and predictors of high LVMD were assessed using multivariable regression.

Results

The ULN for LVMD was 53 ms and LVMD>ULN was present in 11.8% of the full cohort. These individuals were older, more often male, had higher body mass index (BMI) and more frequently had hypertension, hyperlipidaemia and previous myocardial infarction. They also had higher LVMI (57±11 g vs 52±11 g, p<0.001). CACS ≥100 was present in 17.8% vs 10.3% (OR 1.90, 95% CI 1.46 to 2.48, p<0.001) in those with increased versus normal LVMD. Age, BMI, LVMI and heart rate were independently associated with LVMD.

Conclusions

LVMD above ULN was observed in 11.8% of this middle-aged general population sample and was associated with a more adverse cardiovascular risk profile and markers of cardiac remodelling, particularly higher LVMI and larger indexed left ventricular volumes.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Left ventricular mechanical dispersion (LVMD) reflects heterogeneity in left ventricular myocardial contraction, and increased LVMD is associated with ventricular arrhythmias and myocardial disease, but its distribution and clinical correlates in the general population are not well established.

WHAT THIS STUDY ADDS

• In a large population-based cohort of middle-aged individuals, increased LVMD was present in approximately 12% and was associated with a more adverse cardiovascular risk profile and markers of subclinical cardiac remodelling.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• LVMD may serve as a marker for early myocardial alterations in the general population, but further studies are needed to determine its prognostic value and potential role in cardiovascular risk stratification.

Background

Cardiovascular (CV) disease is one of the most common causes of death globally.¹ Early and more precise identification of individuals with an increased risk of myocardial infarction, sudden cardiac death and cardiac remodelling is essential for effective prevention.² There is also an additional need for improved risk stratification among patients already diagnosed with CV disease, where better identification of high-risk individuals may guide clinical decision-making and treatment.³

Heterogeneous left ventricular (LV) myocardial contraction, measured as left ventricular mechanical dispersion (LVMD) by speckle-tracking echocardiography, has been described in several cardiac disease states, including postmyocardial infarction, hypertrophic cardiomyopathy and non-ischaemic dilated cardiomyopathy, where more pronounced LVMD has been associated with ventricular arrhythmias.4,6 LVMD has also been suggested to be a stronger predictor of ventricular arrhythmias than left ventricular ejection fraction (LVEF).⁷

Previous community-based cohort studies have explored LVMD in the general population and reported associations with clinical characteristics, while also providing important reference values.8,10 However, direct comparisons between studies are limited by differences in study size, age distribution and echocardiographic methodology. Consequently, the prevalence and clinical implications of increased LVMD in the general population remain incompletely understood. CV disease often begins to develop during midlife, when preventive interventions may still be effective. Identifying early mechanical abnormalities at this stage could therefore improve risk stratification beyond conventional measures. We hypothesised that increased LVMD would be associated with an adverse CV risk profile and markers of cardiac remodelling in a middle-aged general population. Accordingly, the objectives of this large population-based, cross-sectional study were threefold: (1) to establish an upper limit of normal (ULN) for LVMD in individuals aged 50–64 years, (2) to determine the prevalence of increased LVMD and (3) to examine its associations with CV risk factors and early signs of cardiac remodelling.

Methods

Study population

The Swedish Cardiopulmonary Bioimage Study (SCAPIS) is a population-based cohort study aimed at investigating and ultimately reducing manifestations of CV and pulmonary disease and mortality (www.scapis.org). The study includes approximately 30 000 individuals aged 50–64 years recruited from the general population during 2013–2018. This age interval was selected because cardiometabolic risk factors and subclinical atherosclerosis often start to develop during these years.¹¹ There were no exclusion criteria, except for the inability to comprehend written and spoken Swedish required for the informed consent. The present study is a substudy based on the SCAPIS cohort in Linköping. Individuals aged 50–64 years, living in the Linköping municipality, were randomly selected from the Swedish population register. Invitation letters were sent by mail according to the SCAPIS recruitment procedure, and non-responders were reminded by telephone and letter. In total, 5057 individuals (58% of invited) accepted and underwent investigations in Linköping during 2015–2018.¹² In addition to the SCAPIS core protocol, all participants also underwent a transthoracic echocardiography.¹² Subjects with suboptimal echocardiographic image quality or ECG abnormalities interfering with LVMD analysis (atrial fibrillation or flutter, left or right bundle branch block, intraventricular or bifascicular block, pre-excitation) were excluded. Individuals with self-reported atrial fibrillation or flutter were also excluded. The final study population and the derivation of the low-risk subgroup used to define the ULN of LVMD are shown in figure 1.

Figure 1. BMI, body mass index; CACS, coronary artery calcium score; COPD, chronic obstructive pulmonary disease; LBBB, left bundle branch block; LVMD, left ventricular mechanical dispersion; RBBB, right bundle branch block; SCAPIS, Swedish Cardiopulmonary Bioimage Study; SIS, segmental involvement score.

Anthropometry, ECG and questionnaires

Body weight was measured on a balance scale with participants wearing light clothing and no shoes. Body mass index (BMI) was calculated as body weight (kg)/(body length (m))². Body surface area (BSA) was calculated using the Mosteller formula.¹³ Heart rate was measured during the echocardiographic examination. A 12-lead ECG was recorded after participants rested supine for 5 min. Participants also provided a detailed medical history through questionnaires.¹² Heart failure, hyperlipidaemia, hypertension, myocardial infarction, coronary revascularisation and stroke were defined as doctor-diagnosed, self-reported conditions. Angina was defined as a composite of self-reported diagnosis and affirmative responses to symptom-related questions. Educational level was self-reported and entered as an ordinal variable coded as 0=no completed compulsory education, 1=compulsory school or equivalent, 2=upper secondary school, folk high school or vocational training and 3=university or college degree.

Biochemistry

A 100 mL venous blood sample was collected from each participant for biobanking after an overnight fast. Cholesterol, high-density lipoprotein, triglycerides, low-density lipoprotein, plasma glucose, HbA1c, high-sensitivity C-reactive protein and creatinine were immediately analysed at the Department of Clinical Chemistry, Linköping University, which is accredited according to the standard SS-EN ISO/IEC 17025:2018. Diabetes was defined as a fasting plasma glucose (fP-glucose) ≥7.0 mmol/L and/or HbA1c ≥48 mmol/mol and/or self-reported known diabetes. Pre-diabetes was defined as fP-glucose 6.1–6.9 mmol/L and/or HbA1c 42–47 mmol/mol.

On-site and home blood pressure measurements

Systolic and diastolic blood pressures were measured twice in each arm, and the average of these readings was used. Additionally, participants measured their blood pressure at home two times per day (morning and afternoon) for 1 week, except on day 1 (morning omitted), following a detailed protocol.¹⁴ Each session recorded the mean of three readings taken 1 min apart, with a total of 39 measurements across 13 occasions. The mean of morning measurements was used for this study. All measurements were taken using an automatic device (Omron M10-IT, Omron Health Care, Kyoto, Japan).

Accelerometry

Time spent sedentary was measured using an accelerometer (Actigraph GT3X+, wGT3X+ and wGT3X-BT) worn at the hip for at least 4 days during waking hours. Data were recorded from three different axes and combined into one resulting vector. Data were extracted in 60 s intervals and expressed as counts per minute, where sedentary time was defined as <200 counts per minute. The average sedentary time per day over the 4-day period was calculated and expressed in minutes.¹⁵

Echocardiography

A comprehensive echocardiographic examination was performed by trained and experienced sonographers using a Vivid E95 ultrasound scanner with M5Sc probe (General Electric Healthcare, Chicago, Illinois, USA). The images were stored, and measurements were made offline using the EchoPAC software V.201 (GE Vingmed Ultrasound, General Electric, Chicago, Illinois, USA).

LV inner diameter and septal and posterior wall thickness were measured at end-diastole in the parasternal long-axis view according to current recommendations.^{16 17} LV end-diastolic volume, indexed to BSA, and LVEF were measured by the modified Simpson’s method using automated LVEF in EchoPAC and manually adjusted when needed.

Longitudinal strain was analysed offline in the three apical views (four-chamber, two-chamber and apical long-axis views) using the automated functional imaging (AFI) module in EchoPAC. The region of interest was manually adjusted when needed; strain curves were visually inspected; and segments with failed tracking or obscured walls were excluded. Image quality was considered adequate for analysis if at least 16 myocardial segments were suitable for strain analysis, and LVMD was derived from the automatically identified peak negative longitudinal strain. Global longitudinal strain (GLS) was calculated as the mean of peak longitudinal strain across all available LV segments according to recommendations from the European Association of Cardiovascular Imaging.¹⁶ LVMD was defined as the SD of time-to-peak longitudinal strain (from R-wave onset) across all LV segments and was automatically calculated by AFI (figure 2).

Figure 2. Screen capture of segmental longitudinal strain curves from six left ventricular segments in an apical four-chamber view. The y-axis shows longitudinal strain, and the x-axis represents time during the cardiac cycle. The arrows illustrate differences in the timing of peak systolic strain between segments. AVC, aortic valve closure.

Peak early (E) and late (A) transmitral flow velocities were obtained by pulsed wave Doppler with the sample volume placed at the tips of the mitral valve leaflets, and the E/A ratio was calculated. Early diastolic mitral annular velocity (e′) was measured by tissue Doppler imaging at the septal and lateral annulus, and the average e′ was used to calculate E/e′.

Computed tomography

Coronary artery calcium was determined from ECG-gated non-contrast CT with a Stellar detector (Somatom Definition Flash, Siemens Medical Solutions). Calcifications in the left main/left anterior descending, circumflex and right coronary arteries were identified using the SyngoVia calcium scoring software (Volume Wizard, Siemens Healthineers, Erlangen, Germany) and evaluated according to the 18-coronary segment model defined by the Society of Cardiovascular Computed Tomography.¹⁸ The coronary artery calcium score (CACS) was calculated using Agatston’s method and categorised as 0, 1–99 (mild), 100–399 (moderate) or >400 (high).^{19 20}

Left ventricular mass (LVM) was measured offline on the contrast-enhanced images in late diastole using Segment CT Research (Medviso, Lund, Sweden) via automated endocardial and epicardial delineation. Myocardial tissue volume was multiplied by myocardial tissue density (1.055 g/mL).²¹ CT-based volumetric assessment of LVM is done without geometrical assumptions and has good reproducibility.²² Manual quality control was performed in 5% of cases, and LVM values±3 SDs were excluded.²² LVM was indexed to BSA to render the left ventricular mass index (LVMI).

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics V.29. To define the ULN for LVMD, we identified individuals with the lowest CV risk, excluding those with moderate/severe valve pathology; previous myocardial infarction; coronary artery revascularisation; angina pectoris; heart failure; stroke; hypertension; chronic obstructive pulmonary disease, emphysema or chronic bronchitis; hyperlipidaemia; diabetes; pre-diabetes; CACS >0; segmental involvement score >0; LVEF <50%; GLS <−18%; or BMI >30 (see figure 1). The ULN was set at the 95th percentile for LVMD in this subgroup.

For the full study cohort, characteristics were compared between individuals below or above the ULN for LVMD. Continuous variables were reported as mean±SD if normally distributed or median (IQR) if skewed distribution, and categorical variables were expressed as number (%). Group comparisons were conducted using the independent t-test, Mann-Whitney U test or χ² test depending on the type and distribution of data.

Univariable and multivariable linear regression was used to assess associations between LVMD and CV risk factors in the full cohort. Variables included in the univariable analysis were sex, age, BMI, average morning home systolic blood pressure for 7 days, HbA1c, hypercholesterolaemia, heart rate, time spent sedentary, LVMI from CT, left ventricular end-diastolic volume index (LVEDVI) from echocardiography, GLS, LVEF and CACS. CACS was log transformed before analysis. Variables were selected based on clinical relevance and established CV risk factors, and those with a p value <0.1 in univariable analyses were included in multivariable models. Because GLS and LVEF capture overlapping aspects of systolic function they were not entered simultaneously. An alternative model with LVEF instead of GLS is provided in the online supplemental material.

Univariable and multivariable logistic regression was also performed to investigate risk factors for LVMD>ULN in the full cohort. The same set of predictors was used, except that CACS was entered as a binary variable with a threshold of 100. Additionally, binary logistic regression analyses were performed to assess the associations between LVMD >53 ms and different CACS levels (CACS >1, >100 and >400).

Collinearity was assessed using variance inflation factor and tolerance. A variance inflation factor >5 and tolerance <0.2 were considered problematic. Intraclass correlation coefficients (ICC) were used to assess reproducibility, and values were interpreted as poor if <0.50, moderate if 0.50–0.75, good if 0.75–0.90 and excellent if >0.90. A p value <0.05 was considered statistically significant.

Results

Mechanical dispersion in the general population

Among the 3997 subjects included in the study 51.4% were female and the median age was 57.1 years (IQR 53.4–64.3). Mean BMI was 26.4±3.9, 18.6% had hypertension, 5.8% had diabetes and 12.6% exhibited pre-diabetes or diabetes. Approximately 2% of participants had a history of myocardial infarction, percutaneous coronary intervention or coronary artery bypass grafting. Heart failure was present in 0.3% of the cohort, while moderate or severe valve disease was observed in 1.5%. LVMD was normally distributed and ranged from 8 ms to 103 ms. Mean LVMD in the full sample was 41±11 ms, the median value was 40 ms (33−47). Characteristics for included and excluded participants are shown in online supplemental table S1.

Intraobserver and interobserver reproducibility for LVMD measurements was assessed in 18 subjects. The intraobserver ICC was 0.88 (95% CI 0.71 to 0.95) and the interobserver ICC was 0.88 (95% CI 0.72 to 0.96), indicating good reliability.

ULN for LVMD

To establish a ULN for LVMD, a lower risk subgroup, free from known cardiopulmonary and metabolic diseases and risk factors, was defined as detailed in the Methods section. This group consisted of 1165 individuals, 65.2% female (n=759). Mean age was 56.1±4.2 years. LVMD was normally distributed with a mean value of 38±9 ms, and a range from 10 ms to 88 ms. There was no significant difference in LVMD between sexes (p=0.631). Using the 95th percentile, the ULN for LVMD was 53 ms. Characteristics for this subgroup are shown in online supplemental table S2.

Associations between CV risk factors and LVMD

In the entire cohort, LVMD>ULN was observed in 11.8% of participants. Characteristics of the full cohort, divided into groups with normal LVMD or LVMD>ULN, are presented in tables1 2. In brief, subjects with LVMD>ULN (mean 61±8 ms) were more often male, had higher BMI and had a higher frequency of CV morbidity. They also had greater LVMI on CT, slightly lower GLS and LVEF and higher CACS.

Table 1. Characteristics of individuals with LVMD above or below 53 ms.

	LVMD ≤53 ms n=3527 (88.2%)	LVMD >53 ms n=470 (11.8%)	P value
Age, years	56.8 (53.2–60.9)	59.5 (55.8–62.5)	<0.001
Female	1869 (53.0)	184 (39.1)	<0.001
BMI, kg/m2	26.2±3.9	27.5±3.9	<0.001
BMI >30	521 (14.8)	118 (25.1)	<0.001
Systolic blood pressure, mm Hg	131 (±17)	140 (±19)	<0.001
Diastolic blood pressure, mm Hg	82 (±10)	87 (±11)	<0.001
Hypertension diagnosis	574 (16.7)	151 (32.9)	<0.001
Diabetes	197 (5.6)	35 (7.4)	0.144
Pre-diabetes	420 (11.9)	83 (17.7)	0.002
Myocardial infarction	41 (1.2)	16 (3.5)	<0.001
Coronary revascularisation	31 (0.9)	13 (2.8)	<0.001
Angina pectoris	99 (2.8)	18 (3.8)	0.217
Heart failure	4 (0.1)	6 (1.3)	<0.001
Stroke	39 (1.1)	4 (0.9)	0.61
COPD, chronic bronchitis or emphysema	16 (0.5)	3 (0.7)	0.589
Hyperlipidaemia	291 (8.5)	68 (14.8)	<0.001
Smoking	317 (9.3)	55 (12.1)	0.057
Total cholesterol, mmol/L	5.5±1.1	5.5±1.1	0.364
Triglycerides, mmol/L	1.0 (0.8–1.4)	1.1 (0.9–1.6)	<0.001
HDL-C, mmol/L	1.6 (1.3–2.0)	1.5 (1.2–1.8)	<0.001
LDL-C, mmol/L	3.3±0.9	3.3±1.0	0.868
GFR, mL/min/1.73 m2	93±21	98±22	<0.001
CRP, mg/L	0.9 (0.4–1.7)	1.0 (0.5–2.3)	0.002
Troponin I, ng/L	1.7 (1.3–2.7)	2.1 (1.3–3.4)	<0.001
BNP, pg/mL	48 (28–79)	48 (29–92)	0.053

Values are mean±SD, median (IQR) or absolute number (per cent) depending on the type and distribution of data. Percentages are based on available data for each variable.

BMI, body mass index; BNP, brain natriuretic peptide; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; GFR, glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; LVMD, left ventricular mechanical dispersion.

Table 2. Echocardiographic and CT characteristics of individuals with LVMD above or below 53 ms.

	LVMD ≤53 ms 3527 (88.2%)	LVMD >53 ms 470 (11.8%)	P value
Echocardiography
LVEDV, mL	97±24	107±31	<0.001
LVEDVI, mL/m²	50±10	53±13	<0.001
LVEF, %	60±5	59±6	<0.001
GLS, %	−20.1±2.1	−18.7±2.3	<0.001
LVMD, ms	38±8	61±8	<0.001
E/e′	10.0±3.0	11.2±3.4	<0.001
E/A	1.3±0.4	1.1±0.3	<0.001
CT angiography
LVM, g	100.8±26.1	114.1±27.3	<0.001
LVMI, g/m2	52.1±10.6	56.5±10.8	<0.001
CACS 0	2154 (62.5)	215 (48.0)	<0.001
CACS 1–99	938 (27.2)	153 (34.2)
CACS 100–399	248 (7.2)	53 (11.8)
CACS >400	106 (3.1)	27 (6.0)
CT-detected stenosis >50%	128 (3.8)	21 (4.7)	0.351

Values are mean±SD, median (IQR) or absolute number (per cent) depending on the type and distribution of data. Percentages are based on available data for each variable.

A, transmitral A-wave velocity; CACS, coronary artery calcium score; e′, early diastolic mitral annulus velocity; E, transmitral E-wave velocity; GLS, global longitudinal strain; LVEDV, left ventricular end-diastolic volume; LVEDVI, left ventricular end-diastolic volume index; LVEF, left ventricular ejection fraction; LVM, left ventricular mass; LVMD, left ventricular mechanical dispersion; LVMI, left ventricular mass index.

In univariable linear regression, sex, age, BMI, HbA1c, hypercholesterolaemia, heart rate, LVMI, CACS, LVEDVI and GLS were significantly associated with LVMD. In the multivariable model, sex, age, BMI, heart rate, LVMI, LVEDVI and GLS remained independently associated with LVMD (table 3). A sensitivity analysis replacing GLS with LVEF yielded similar results (online supplemental table S3).

Table 3. Associations between cardiovascular risk factors and LVMD in univariable and multivariable linear regression.

	Univariable linear regression				Multivariable linear regression
	Standardised coefficients Beta	Unstandardised coefficients B	95% CI for B	P value	Standardised coefficients Beta	Unstandardised coefficients B	95% CI for B	P value
Sex*	0.040	0.893	0.20 to 1.56	0.011	−0.156	−3.479	−4.34 to −2.62	<0.001
Age	0.144	0.366	0.29 to 0.44	<0.001	0.142	0.363	0.28 to 0.45	<0.001
BMI	0.107	0.305	0.22 to 0.39	<0.001	0.118	0.343	0.25 to 0.44	<0.001
Systolic blood pressure†	−0.002	−0.002	−0.03 to 0.02	0.886
HbA1c	0.050	0.087	0.03 to 0.14	0.002	0.002	0.004	−0.06 to 0.07	0.894
Hypercholesterolaemia	0.048	1.838	0.62 to 3.05	0.003	0.021	0.840	−0.48 to 2.16	0.212
Heart rate	−0.120	−0.129	−0.16 to −0.10	<0.001	−0.138	−0.147	−0.18 to −0.11	<0.001
Time spent sedentary	−0.023	−0.003	−0.006 to 0.001	0.155
LVMI	0.149	0.155	0.11 to 0.15	<0.001	0.115	0.120	0.08 to 0.16	<0.001
lnCACS	0.083	0.452	0.28 to 0.62	<0.001	0.020	0.111	−0.08 to 0.30	0.247
LVEDVI	0.106	0.113	0.08 to 0.15	<0.001	0.072	0.078	0.04 to 0.12	<0.001
GLS	0.264	1.361	1.21 to 1.52	<0.001	0.249	1.309	1.14 to 1.48	<0.001

Bold numbers indicate statistically significant associations (p<0.05). Dependent variable: LVMD (ms). Positive B indicates higher LVMD.

^*Male=1, female=0.

^†Systolic blood pressure is the mean of six values from morning home measurements.

BMI, body mass index; GLS, global longitudinal strain; HbA1c, glycosylated haemoglobin; lnCACS, natural log-transformed coronary artery calcium score; LVEDVI, left ventricular end-diastolic volume index; LVMD, left ventricular mechanical dispersion; LVMI, left ventricular mass index (from CT).

In multivariable logistic regression, including GLS, LVMD>ULN (>53 ms) was independently associated with age, BMI, heart rate, GLS, LVMI and LVEDVI, whereas sex, HbA1c and CACS >100 were not significant contributors to the model (online supplemental tables S4 and S5). A sensitivity analysis replacing GLS with LVEF yielded similar results (online supplemental table S6).

Adjustment for educational level did not alter the main results, and educational level was not independently associated with LVMD (online supplemental table S7).

Additional analyses showed that CT-detected coronary stenosis >50% was not independently associated with LVMD in an alternative multivariable model in which stenosis >50% replaced lnCACS (online supplemental table S8).

No relevant multicollinearity was detected in either model. All predictors showed a variance inflation factor <2 and tolerance >0.65.

In a further binary logistic regression analysis with LVMD >53 ms as the dependent variable, increasing CACS thresholds were significantly associated with higher odds of elevated LVMD (online supplemental table S9). For individuals with CACS >100, the OR was 1.9 (95% CI 1.46 to 2.48, p<0.001). For CACS >400, the OR increased to 2.0 (95% CI 1.31 to 3.12, p=0.002).

Inter-relationship between LVEF, GLS and LVMD

When LVEF, GLS and LVMD were compared, 2868 individuals (74.8%) had values within normal limits across all three measures (LVEF ≥50%, GLS ≥−18% and LVMD ≤53 ms), whereas 964 individuals (25.2%) had an abnormal value in at least one measure (figure 3). Only 20 individuals (0.5%) had abnormal values in all three measures. The largest overlap was observed between increased LVMD and reduced GLS, with 142 individuals (3.7%) exhibiting both abnormalities. Isolated abnormalities were common, particularly for GLS and LVMD, with isolated reduced GLS in 433 individuals (11.3%) and isolated increased LVMD in 286 individuals (7.5%).

Figure 3. Panel A shows a Venn diagram illustrating the relationship between reduced LVEF (<50%), reduced GLS (<−18%) and prolonged LVMD (>53 ms) among subjects with available data for all three measures and at least one abnormal value (n=964). The area of each circle in panel A is proportional to the number of subjects with the respective abnormal echocardiographic variable, whereas the overlapping areas are approximate. Panel B shows a scatter plot of GLS versus LVMD. GLS, global longitudinal strain; LVEF, left ventricular ejection fraction; LVMD, left ventricular mechanical dispersion.

Discussion

The most important finding of this population-based study is that LVMD>ULN is associated with an adverse CV risk profile and imaging markers of subclinical cardiac remodelling. For those aged 50–64 years, we propose a ULN for LVMD of 53 ms.

LVMD has previously been linked to ventricular arrhythmias, cardiomyopathies and diastolic dysfunction.423,25 In this general population sample, individuals with LVMD>ULN exhibited mean LVMD values previously associated with adverse outcomes in patients with ischaemic heart disease or cardiomyopathy.⁷ The mechanisms underlying prolonged LVMD remain incompletely understood, but fibrosis is thought to play a key role. Haland et al demonstrated in a study involving patients with hypertrophic cardiomyopathy that the extent of LVMD correlated with the degree of myocardial fibrosis, as detected by cardiac MRI.⁴ Similarly, LVMD has been linked to myocardial scar burden, indicating that regions of scar tissue contribute to mechanical dispersion.^{26 27} While fibrosis plays a crucial role in prolonged LVMD, it is unlikely to fully explain the phenomenon. Electrical dyssynchrony, such as that seen in left bundle branch block, is likely another contributing factor.²³ Increased LVMD may therefore in some cases reflect conduction delay rather than myocardial structural alterations, which is why subjects with major ECG abnormalities expected to interfere with LVMD analysis were excluded from this study.

Evidence also suggests that LVMD exhibits some degree of plasticity, indicating that additional factors may be involved. Hasselberg et al showed a significant reduction in LVMD 6 months after cardiac resynchronisation therapy, with the greatest improvement among patients who also experienced reductions in ventricular volume.²⁴ Thus, while fibrosis is an important factor in prolonged LVMD, other mechanisms, including electrical dyssynchrony, ventricular geometry and functional remodelling, are most likely also involved.

Heart rate was associated with LVMD, with higher heart rates correlating with lower LVMD values. As LVMD is expressed in absolute time, a shorter systolic duration at higher heart rates leaves less time for dispersion in the contraction. LVMD should therefore be interpreted in relation to heart rate. Although LVMD is most commonly reported in absolute milliseconds, as in the present study, normalisation to RR interval has been used in some studies to express mechanical dispersion as a proportion of the cardiac cycle.²⁸ Whether RR-normalised LVMD may provide incremental clinical value compared with absolute LVMD remains to be established.

Another key finding in our study was the consistent association between LVMI, LVEDVI and LVMD across all analyses. Although differences in LV size and mass were modest and within normal reference ranges, LVMD>ULN was already observed, supporting the concept that mechanical dispersion is associated with subtle myocardial alterations even in the absence of overt structural remodelling. Given that increased LVMI is a well-established CV risk factor linked to higher mortality, this relationship is particularly relevant.29,31 While hypertension is a major contributor to LV hypertrophy it was not independently associated with LVMD in our models.³² This aligns with previous findings suggesting that increased LVMD primarily reflects secondary myocardial alterations, such as early fibrosis and remodelling, rather than hypertension itself.⁹ These associations persisted after adjustment for clinical risk factors and systolic function (LVEF and GLS), indicating that LVMD>ULN is associated with structural remodelling beyond what is captured by other measures of systolic function. Together, these findings suggest that LVMD>ULN is associated with subtle myocardial changes in a general population, the clinical and prognostic implications of which require confirmation in longitudinal studies.

Although LVMD was associated with GLS and LVEF, most subjects with LVMD>ULN had normal LVEF and GLS. This suggests that LVMD is not interchangeable with these systolic function markers but provides complementary information. LVEF, though widely used, is an imprecise prognostic marker, particularly when only mildly reduced, whereas GLS has been shown to be superior.^{33 34} In selected patient populations LVMD has been demonstrated to be associated with ventricular arrhythmias to a greater extent than both LVEF and GLS and whether it may convey further prognostic information in general population settings remains to be determined.⁷

While CACS was not independently associated with LVMD in the multivariable models, significant associations were observed in univariate analyses and in logistic regressions across increasing CACS categories. These findings suggest that LVMD>ULN more frequently co-occurs with higher CACS levels in this population but does not reflect an independent relationship. Given that CACS is closely linked to age and cardiometabolic risk factors the observed associations likely reflect shared underlying risk factors rather than an independent relationship between coronary atherosclerosis and LVMD.³⁵

There has been some prior work aimed at establishing reference values. Cheng et al investigated the normal distribution for LVMD in a Framingham cohort of 738 individuals but used a different method with only 12 myocardial segments.³⁶ Rodríguez-Zanella et al reported a ULN of 64 ms for individuals aged 51 and above, though their study was limited by a relatively small sample size.¹⁰ The largest previous study by Aagaard et al reported a ULN of 61 ms in 64 year-olds. In our study, the lower ULN likely reflects inclusion of younger participants and stricter reference criteria.⁸ The size of our study population, combined with its extensive characterisation, provides reliable data on a ULN of LVMD for the included age span.

Strengths and limitations

This study benefits from the large and well-characterised SCAPIS cohort which was examined by CT of the heart and coronary arteries, echocardiography, home blood pressure measurements and physical activity assessment. LVM was measured from cardiac CT, which through volumetric segmentation provides higher reliability than echocardiographic measurements.²² Interfering elements were reduced by excluding people with cardiac arrhythmia, left and right bundle branch blocks and pacemakers, thus the results cannot be extrapolated to populations with arrhythmias or conduction abnormalities. The relatively narrow age range limits the generalisability but importantly represents a critical period for CV disease onset and prevention.

Residual confounding by, for instance, diet, alcohol consumption and sleep patterns cannot be excluded. Selection bias may also have been present, as SCAPIS participants had somewhat higher socioeconomic status and were more likely to reside in affluent areas than the target population.³⁷ This may have affected generalisability, although the effect on most baseline cardiometabolic risk factors appears to have been small.³⁷

As with all strain measurements, imaging quality was occasionally suboptimal. LVMD measurements with good data quality were nevertheless available for 80% of the cohort. Participants with suboptimal image quality tended to have a less favourable risk profile, and it is possible that factors such as higher BMI and smoking history have impaired echocardiographic imaging. This may have contributed to some differences between included and excluded participants.

Finally, as a cross-sectional study, conclusions on causality and a definite link between LVMD and adverse CV events cannot be established. However, as the SCAPIS study continues, longitudinal data may allow investigation of the prognostic impact of increased LVMD in the future.

Conclusions

A high degree of myocardial contraction heterogeneity was present in approximately 12% of this general population sample and was associated with a more severe CV risk profile and subclinical cardiac remodelling. The clinical utility of LVMD is not yet fully established, but it may reflect subtle myocardial alterations. Future studies are needed to clarify whether increased LVMD is associated with a higher risk of adverse CV outcomes in the general population.

Data availability statement

Data may be obtained from a third party and are not publicly available.