Objectives Speckle-tracking echocardiography enables detection of abnormalities in cardiac mechanics with higher sensitivity than conventional measures of left ventricular (LV) dysfunction and may provide insight into the pathogenesis of coronary heart disease (CHD). We investigated the relationship of LV longitudinal strain, LV early diastolic strain rate (SR) and left atrial (LA) reservoir strain with long-term CHD incidence in community-dwelling older adults.
Methods The association of all three strain measures with incidence of non-fatal and fatal CHD (primary outcome of revascularisation, non-fatal and fatal myocardial infarction) was examined in the population-based Cardiovascular Health Study using multivariable Cox proportional hazards models. Follow-up was truncated at 10 years.
Results We included 3313 participants (mean (SD) age 72.6 (5.5) years). During a median follow-up of 10.0 (25th–75th percentile 7.7–10.0) years, 439 CHD events occurred. LV longitudinal strain (HR=1.25 per SD decrement, 95% CI 1.09 to 1.43) and LV early diastolic SR (HR=1.31 per SD decrement, 95% CI 1.14 to 1.50) were associated with a significantly greater risk of incident CHD after adjustment for potential confounders. By contrast, LA reservoir strain was not associated with incident CHD (HR=1.06 per SD decrement, 95% CI 0.94 to 1.19). Additional adjustment for biochemical and echocardiographic measures of myocardial stress, dysfunction and remodelling did not meaningfully alter these associations.
Conclusion We found an association between echocardiographic measures of subclinically altered LV mechanics and incident CHD. These findings inform the underlying biology of subclinical LV dysfunction and CHD. Early detection of asymptomatic myocardial dysfunction may offer an opportunity for prevention and early intervention.
- coronary artery disease
Data availability statement
Cardiovascular Health Study data are available upon request (https://biolincc.nhlbi.nih.gov/studies/chs/).
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Despite recent declines in its prevalence, coronary heart disease (CHD) is the leading cause of death in the USA.1 Early detection of asymptomatic myocardial dysfunction may offer an opportunity for prevention and early intervention. Speckle-tracking echocardiography (STE) detects abnormalities in myocardial performance with higher sensitivity than conventional measures of left ventricular (LV) dysfunction.2 3 LV longitudinal strain is an indicator of longitudinal (subendocardial) fibre systolic function, while LV early diastolic strain rate (SR) represents the rate of myocardial deformation in early diastole and is a sensitive measure of longitudinal fibre relaxation. The subendocardial layer of the LV is the most susceptible to ischaemia; thus, LV strain parameters could be a sensitive measure of subclinical CHD.4 Left atrial (LA) reservoir strain is a measure of the ability of the LA to fill during LV systole.5 Abnormalities in LA reservoir strain are indicative of elevated LA pressure and reduced LA compliance, both of which can occur in the setting of CHD.6
At a population level, LV longitudinal strain was found to be associated with incident cardiovascular events in prior studies,7–9 which included largely middle-aged cohorts with relatively small sample sizes and/or heterogeneous composite outcomes and did not focus on older adults, a population with a high prevalence of subclinical cardiovascular disease.10 Moreover, prior studies have not concurrently examined strain measures of LV diastolic relaxation or LA function. In view of these gaps, we undertook speckle-tracking measurement of indices of LV and LA mechanics on digitised echocardiograms in the Cardiovascular Health Study (CHS). We tested the hypothesis that LV longitudinal strain, LV early diastolic SR and LA reservoir strain are each associated with non-fatal and fatal CHD (primary outcome of revascularisation, non-fatal and fatal myocardial infarction (MI)) independent of clinical and laboratory covariates in CHS.
CHS is a prospective, population-based study of community-dwelling, ambulatory adults ≥65 years old; eligible participants were sampled from Medicare eligibility lists at four US field centres, as previously described.11 An original cohort of 5201 mostly white participants were recruited in 1989–1990, followed by a supplemental cohort of 687 predominantly African American individuals in 1992–1993. Inperson examinations were performed annually through 1998–1999. Telephone interviews were conducted annually at a 6-month offset from the inperson visits from 1989 to 1999 and semiannually thereafter.12 13 For this analysis, participants who had a history of CHD, heart failure, stroke/transient ischaemic attack and atrial fibrillation at the time of echocardiogram (for the original cohort) or prior to their echocardiogram (for the supplemental cohort) were excluded. Only participants who had echocardiograms with speckle-tracking strain analysis of either LA or LV were included. Participants were not involved in design/conduct or reporting/dissemination of this research.
Comprehensive two-dimensional, M-mode and Doppler echocardiograms were performed at the time of enrolment using a standardised protocol and core reading centres, as previously described.14 Images were recorded on Super VHS (video home system) tapes using Toshiba SSH-160A ultrasound machines at the four field centres. Initial measurements were made at the central core laboratory (Baltimore, Maryland).
Archived echocardiograms were digitised (2016–2018) using the TIMS 2000 DICOM System (Foresight Imaging, Chelmsford, Massachusetts) with methods developed for the Hypertension Genetic Epidemiology Study.15 16 Cine loops of two to four cardiac cycles from the apical four-chamber view were digitised at a rate of 30 frames per second (fps). STE analyses were performed by five experienced readers (TOMTEC V.4.5, Unterschleiβheim, Germany). All echocardiograms were assigned chamber-specific image quality scores (0–4).15 Electrocardiographic gating was used to define the cardiac cycle. The LV endocardial border was manually traced in the apical four-chamber view. Six LV segments were identified by the software. Segments that did not track appropriately were removed from the analysis. The average of the remaining segments was generated, and LV longitudinal strain and LV early diastolic SR were derived. LA reservoir strain was defined as the peak average LA strain in the apical four-chamber view. Reproducibility of strain measures is shown in online supplemental table 1. All strain parameters were reported as absolute values for ease of display and interpretation of results. Septal speckle-tracking e’ (STe’) velocities were measured from STE of the basal septum, which has been validated against conventional tissue Doppler imaging (TDI) but results in lower values than with TDI.15
LA diameter was measured in the parasternal long-axis view. LA volume was calculated using the estimated ellipsoid method in the apical four-chamber view.17 LV mass was derived using the Devereux formula.18
The primary outcome of this study was time to first non-fatal and fatal CHD (primary outcome of percutaneous angioplasty, coronary artery bypass surgery, non-fatal and fatal MI), identified using International Classification of Diseases-9 codes from hospitalisation records that were prospectively gathered from all CHS participants every 6 months and centrally adjudicated, as previously described.12 19 Follow-up for incident CHD was complete and extended from baseline visit at the time of echocardiogram until incident CHD event or death. Follow-up was truncated at 10 years. Additionally, the three components of the primary outcome were analysed separately.
All covariates were collected at baseline visit at the time of echocardiogram, unless otherwise noted. Information on lifestyle habits, anthropometry and blood pressure was obtained using standardised methods.11 Prevalent diabetes mellitus was defined by fasting glucose ≥126 mg/dL, non-fasting glucose ≥200 mg/dL or treatment with hypoglycaemic medication. Low-density lipoprotein (LDL) cholesterol, estimated glomerular filtration rate (eGFR) (based on cystatin C) and C reactive protein (CRP) were calculated/measured as previously described.20–22 CHD, heart failure, atrial fibrillation, and stroke or transient ischaemic attack were ascertained at enrolment for the original (1989–1990) and supplemental (1992–1993) cohorts and the intervening period for the supplemental cohort, as previously described.12 23 24 Serum levels of N-terminal pro-B-type natriuretic peptide (NT-proBNP) and high-sensitivity troponin T (hsTnT) were measured in stored specimens collected at baseline examination (Elecsys 2010 analyser, Roche Diagnostics, Indianapolis, Indiana).
Distributions of LV longitudinal strain, LV early diastolic SR and LA strain were assessed by visual inspection of histograms and Q-Q plots and were deemed to be approximately normally distributed. Continuous variables are presented as mean (SD) or median (IQR); categorical variables are presented as counts and percentages. Using Pearson correlation coefficients, we analysed the correlations of the indices of cardiac mechanics with other characteristics. Kaplan-Meier method was used to describe the associations of quartiles of the exposure variables and incident CHD.
Next, the shape of the relationship between each STE parameter and the primary outcome was evaluated, and each STE parameter was modelled continuously (per SD worsening) in Cox proportional hazard models. Confounders were selected a priori for adjustment in sequential models based on prior associations or known biological mechanisms. In the models, we included all participants for whom data were available. The degree of missingness for each variable is described in tables 1 and 2 and in online supplemental table 2. Model 0 included age, race, image quality, field centre and speckle-tracking analyst. In model 1, our main model, we additionally included body mass index (BMI), smoking status, alcohol consumption, oestrogen replacement therapy, systolic blood pressure, treatment with antihypertensive medications, diabetes, LDL, high-density lipoprotein, triglycerides and eGFR. In additional models, we assessed the impact on the risk estimates of adding serum biomarkers of inflammation, myocardial stress and injury (high-sensitivity CRP, hsTnT, NT-proBNP), and separately with the addition of echocardiographic biomarkers of myocardial remodelling and diastolic dysfunction (left ventricular ejection fraction (LVEF), LV mass, STe’, E/STe’ ratio and LA volume) to model 1. We repeated our multivariable Cox regression analyses with STe’ (which is akin to tissue Doppler e’ velocity obtained on standard echocardiograms) as the exposure variable in order to compare it with the three indices of cardiac mechanics and their association with incident CHD. Thereafter, we adjusted the LV strain models for LA reservoir strain. We also included both LV longitudinal strain and LV early diastolic SR in the same model to determine whether one of these parameters was more closely associated with incident CHD.
We repeated the aforementioned multivariable models using one-unit change (instead of 1 SD change) for each of the predictor variables. In sensitivity analyses, we repeated the multivariable models excluding those with reduced LVEF and with abnormal wall motion. We also conducted a complete case analysis. Lastly, we performed an interaction analysis for indices of cardiac mechanics and age, race and sex.
The proportionality assumption was tested using Schoenfeld residuals. All analyses were performed using R V.3.5.1. A two-tailed p<0.05 was used to define statistical significance.
Characteristics of the study cohort and relationships with measures of cardiac mechanics
Of the 5888 CHS participants, 4477 had adequate echocardiograms for speckle-tracking analysis of either the LV or LA. A total of 1164 participants had CHD, heart failure, stroke/transient ischaemic attack or atrial fibrillation at baseline and were excluded. The final study cohort included 3313 participants, with a median (IQR) age of 71.0 (68.0–76.0) years, 61% women and 9.3% African American. The characteristics of participants excluded from the final cohort are displayed in online supplemental table 2. Those who were excluded were older, more likely to be male and black, had a higher prevalence of diabetes and hypertension, higher BMI, greater CRP, hsTnT and NT-proBNP levels, higher LA volumes, and lower eGFR.
The baseline characteristics and correlations with cardiac mechanics indices are displayed in table 1. There was a strong correlation between LV longitudinal strain and LV diastolic SR (r=0.70, p<0.001) and a modest correlation between those measures and LA reservoir strain (r=0.26 and r=0.15, respectively, p<0.001).
Cardiac mechanics and incident CHD
The proportionality assumption was tested using Schoenfeld residuals, which revealed no violations. All variance inflation factors were <3.6, indicating lack of multicollinearity. During a median (IQR) follow-up of 10.0 (7.7–10.0) years, 439 participants had a CHD event (13.3 cases per 1000 person-years); there were 334 incident MIs, 116 percutaneous angioplasties and 149 coronary artery bypass surgeries. For all three echocardiographic measures of cardiac mechanics, there were no significant departures from a linear relationship in their adjusted relationships with incident CHD (figure 1, online supplemental figure 1). In a Kaplan-Meier analysis, incident CHD risk increased with each quartile of worse (lower) LV longitudinal strain, as well as LV early diastolic SR, but not LA reservoir strain (figure 2). In Cox regression models, LV longitudinal strain was associated with incident CHD, adjusted HR 1.25 per SD decrease (95% CI 1.09 to 1.43, p<0.01). Further adjustment for serum biomarkers of myocardial injury and wall stress, as well as standard echo measures of subclinical cardiac disease or LA reservoir strain separately, minimally attenuated the HRs.
Similarly, each SD decrease in LV early diastolic SR was associated with increased hazard for CHD (HR=1.31, 95% CI 0.14 to 0.50, p<0.01). This association remained statistically significant after adjustment for serum and standard echocardiographic biomarkers of myocardial disease, as well as for LA reservoir strain. For both measures of LV myocardial mechanics, these associations were not meaningfully altered after adjustment for the respective other measure (table 2).
LA reservoir strain was not associated with incident CHD (HR=1.06, 95% CI 0.94 to 1.19). Furthermore, STe’ was not associated with incident CHD. In a complete case analysis (online supplemental table 3), the associations remained consistent. Only in the case of LV longitudinal strain additional adjustment for standard echocardiographic measures of subclinical cardiac disease, as well as LA reservoir strain, mildly attenuated the association with incident CHD. There were no significant interactions between any of the STE indices and age, race or sex (online supplemental table 4). In addition, we examined the associations of the three components of the primary outcome separately (online supplemental tables 5.1–5.3). For incident MI and coronary artery bypass surgery, results were consistent with the composite outcome. However, there was no significant association between measures of cardiac mechanics and percutaneous angioplasty (online supplemental table 5.2). Furthermore, we present associations per one-unit change (online supplemental table 6) and among participants with normal LVEF (online supplemental table 7) and normal LV wall motion (online supplemental table 8), which are largely consistent with the main results.
We examined the relationship of three important echocardiographic biomarkers of cardiac mechanics with incident CHD events in older community-dwelling adults. Worse LV longitudinal strain and LV diastolic SR were significantly associated with incident CHD after adjustment for demographic and clinical variables, as well as serum biomarkers and echocardiographic measures of myocardial injury and wall stress. In contrast, LA reservoir strain was not associated with incident CHD. Our findings substantiate the importance of these subclinical indices of LV myocardial mechanics as measures of myocardial health among older adults. Our findings on multivariable Cox regression analysis persisted in the subset of participants with normal LVEF and in those with normal LV wall motion; these results further support the notion that subclinical indices of LV systolic and diastolic longitudinal myocardial mechanics are independently associated with incident CHD in community-dwelling older adults.
Only three prior population-based studies have examined the relationship of speckle-tracking measures of cardiac mechanics with incident CHD events. The findings of our study, encompassing the largest population with the highest number of events, are largely consistent with previous results. However, there was substantial variation in the definition of cardiovascular disease outcomes, which renders direct comparisons difficult. In the Northern Manhattan Study (708 participants, mean age 71 years, 5-year follow-up),7 abnormal longitudinal strain was associated with incident CHD event, despite a low number of events (n=58). In the Copenhagen City Heart Study (1296 participants, mean age 57 years, 11-year follow-up, 149 events),9 LV longitudinal strain was associated with the composite endpoint and with acute MI alone. In the Framingham Offspring Study (2831 participants, mean age 66 years, 6-year follow-up, 69 CHD events8), longitudinal strain was not associated with incident CHD (which included angina pectoris, in contrast to our study). However, in analyses that were unadjusted for multiple testing, LV longitudinal strain was associated with incident CHD.
Interpretation of study findings
While prior studies examined the relationship of longitudinal strain with CHD, our study is the first to focus on adults ≥65 years, the group at highest risk of CHD. Moreover, our study is the first to investigate LV longitudinal strain in parallel with LV early diastolic SR and to demonstrate a significant association of both measures with incident CHD. In fact, both measures were statistically significant when adjusted for each other, suggesting that, despite their high correlation, they provide complementary information. In contrast, e’ velocity, the most commonly used parameter of diastolic function in clinical practice, was not. This may be due to the fact that it is only representative of the LV myocardium near the mitral annulus, whereas early diastolic SR encompasses relaxation of the entire LV, and distal/apical segments may be more affected by CHD than basal segments. Furthermore, LA reservoir strain was not associated with incident CHD, potentially because it may be a less relevant measure for CHD. Increases in LA pressures and alteration in LA mechanics lie further downstream in the cascade of CHD; therefore, LA mechanics may be more relevant to prevalent CHD.
STE enables quantification of myocardial mechanics with greater sensitivity than LVEF alone. Our findings suggest that STE may allow earlier detection of important subclinical LV dysfunction that leads to CHD, upstream of changes in conventional measures in contractility and relaxation in the pathophysiological cascade of ischaemic heart disease. LV longitudinal strain may be particularly sensitive at detecting changes in the subendocardium, which is composed of longitudinally oriented fibres that are especially vulnerable to ischaemia and necrosis. In addition, alterations in myocardial mechanics may denote microvascular dysfunction and chronic ischaemia, which may be amenable to preventative therapy.
Strengths and limitations
To our knowledge, this study is the first to investigate the association of both LV longitudinal systolic and diastolic strain, along with LA reservoir strain, with incident CHD in a population-based cohort of older adults. There are several limitations that deserve consideration. First, LV longitudinal strain was obtained from the apical four-chamber view alone, in contrast to other studies using global longitudinal strain, calculated from three apical views. Given the large number of participants, only one representative view was chosen for these analyses for feasibility purposes. We have previously shown that in population-based studies of participants free of overt LV myocardial disease, LV longitudinal strain in the apical four-chamber view is sufficient to characterise the entire LV.15 In addition, any loss of information from not including the apical two-chamber and three-chamber views would have attenuated the associations we identified. Furthermore, other dimensions of LV strain, for example, circumferential or radial strain, were not obtained. However, longitudinal strain is considered the most sensitive measure of subendocardial function (the layer most susceptible to ischaemia) because it is aligned with the orientation of myocardial fibres. Second, a large proportion of the cohort were excluded owing to suboptimal image quality. This was necessary to maintain the high quality of STE analyses. We contrasted participants that were included and excluded and found only small differences in baseline characteristics, which suggests that inclusion of the entire cohort may not have meaningfully altered the results. Moreover, we adjusted for image quality, field centre and speckle-tracking analyst and would expect variations in image quality to attenuate associations with the outcome. Third, the longitudinal strain values are lower than reported in the clinical literature, largely owing to the digital conversion of VHS videotapes which resulted in a frame rate of 30 fps compared with 60 fps used for direct digital acquisition, potentially leading to lower strain values. The lower values may also be attributable to the TOMTEC software which was used in our analysis. The strain values in this study are not intended to be compared with studies done with different methodology or to be used for clinical diagnosis. However, for a well-powered population health association study as is the case here, the methods used for analysis of archived tapes have been well validated for research purposes.15
Among older community-dwelling adults, LV longitudinal strain and LV early diastolic SR were associated with incident CHD after adjustment for potential confounders, serum and echocardiographic measures of myocardial injury, stress and remodelling. Further work is necessary to understand how this association can be harnessed to improve prediction of CHD events in the individual patient.
What is already known on this subject?
Speckle-tracking echocardiography enables detection of abnormalities in cardiac mechanics with higher sensitivity than conventional measures of left ventricular (LV) dysfunction and may provide insight into the pathogenesis of coronary heart disease (CHD).
LV longitudinal strain and early diastolic strain rate are subclinical indices of LV mechanics and reflect myocardial health.
What might this study add?
Both indices of cardiac mechanics were associated with incident CHD events in the Cardiovascular Health Study, a large population-based study of older community-dwelling individuals.
How might this impact on clinical practice?
The association of LV systolic and diastolic mechanics with incident CHD suggests that LV longitudinal strain parameters may be particularly sensitive at detecting changes in areas that are especially vulnerable to ischaemia and necrosis.
Alterations in myocardial mechanics may denote microvascular dysfunction and chronic ischaemia, which may be amenable to preventative therapy.
These findings inform the underlying biology of subclinical LV dysfunction and CHD.
Data availability statement
Cardiovascular Health Study data are available upon request (https://biolincc.nhlbi.nih.gov/studies/chs/).
Patient consent for publication
This study complied with the Declaration of Helsinki. All field centres (University of Pittsburgh, Johns Hopkins University, Wake Forest University and UC Davis) and the CHS Coordinating Center (University of Washington) received institutional review board approval for the study and all participants gave informed consent.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Twitter @danmassera, @HFpEF
Contributors Conception, design, analysis and interpretation of data: DM, MH, JAD, TB, BMP, JSG, JRK and SJS. All authors participated in drafting of the manuscript or revising it critically for important intellectual content. Final approval of the manuscript submitted was granted by all authors.
Funding This study was funded by R01 HL107577 from the National Heart, Lung, and Blood Institute (NHLBI) to SJS. This research was also supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086 and 75N92021D00006, and grants U01HL080295 and U01HL130114, from NHLBI, with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. JRK was supported by K24HL135413 from NHLBI.
Competing interests DM has received consulting fees from Bristol Myers Squibb. JRK reports stock ownership in Bristol Myers Squibb, Johnson & Johnson, Medtronic, Merck and Pfizer. SJS has received research grants from Actelion, AstraZeneca, Corvia, Novartis and Pfizer; and has received consulting fees from Abbott, Actelion, AstraZeneca, Amgen, Axon Therapies, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, Cardiora, CVRx, Cytokinetics, Eidos, Eisai, GSK, Ionis, Ironwood, Lilly, Merck, MyoKardia, Novartis, Novo Nordisk, Pfizer, Prothena, Sanofi, Shifamed, Tenax and United Therapeutics. All other authors declare that they have no conflicts of interest.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.