Novel cardiovascular magnetic resonance strain heterogeneity phenotypes predict cardiovascular events: A prospective UK Biobank study

Figures

• 

  Fig. 1. Strain curves in ischemic heart disease versus healthy controls. For example, anterolateral MI leads to hypokinesis of the respective LV wall, impairing regional strain and therefore increasing strain CoV without necessarily increasing the SD of time to peak strain i.e. mechanical dispersion. SD standard deviation, MI myocardial infarction.

• 

  Fig. 2. Probability density charts of global strain and strain CoV by CV risk factors and disease. A) Age >65 years, B) Male sex, C) White ethnicity, D) Current smoker, E) Obesity, F) Dyslipidemia, G) Hypertension, H) Diabetes mellitus, I) Chronic kidney disease, J) Coronary artery disease, K) Previous myocardial infarction, L) Stroke, M) Atrial fibrillation and N) Heart failure. CoV coefficient of variation, CV cardiovascular

• 

  Fig. 3. Forest plots, comparing CoV and global strain to predict CV endpoints. CoVCS, CoVRS and GLS were predictive of death from any cause and a composite CV endpoint (MI, HF and arrhythmia) after adjusting for clinical and cardiovascular magnetic resonance imaging risk markers for CV disease. Correction for multiple testing rendered the association between CoVRS, CoVCS, and death from any cause non-significant as well as between CoVRS, HF and arrhythmia non-significant. GCS and GRS were associated with HF. *Filled vs. unfilled marker indicates Bonferroni-corrected P < 0.0033 vs. > 0.0033. CoV coefficient of variation, GLS global longitudinal strain, CoVLS longitudinal strain coefficient of variation, CV cardiovascular, GCS global circumferential strain, CoVCS circumferential strain coefficient of variation, GRS global radial strain, CoVRS radial strain coefficient of variation, MI myocardial infarction, HF heart failure.

• 

  Fig. 4. RCS plots, assessing the non-linear relationship between strain CoV and the outcomes A) composite CV endpoint, B) MI, C) HF, D) arrhythmia and E) death from any cause. There was no graphical or statistical evidence to support a plausible non-linear relationship between strain CoVLS, CoVCS, CoVRS and predefined study endpoints (P > 0.05 for all). LRT Likelihood-ratio test, RCS restricted cubic splines, CoV coefficient of variation, CV cardiovascular, MI myocardial infarction, HF heart failure, CoVLS longitudinal strain coefficient of variation, CoVCS circumferential strain coefficient of variation, CoVRS radial strain coefficient of variation.

• 

  Fig. 5. Strain biomarker KM hazard curves by tertiles for the composite CV endpoint. In participants with reduced LVEF < 50% (panel A), CoVCS and CoVRS stratified the risk of the composite CV endpoint better than global strain. Conversely, in participants with preserved LVEF ≥ 50% (Panel B), GLS, GCS and GRS were more sensitive markers for increased risk of the composite CV endpoint. LVEF left ventricular ejection fraction, CoVLS longitudinal strain coefficient of variation, GLS global longitudinal strain, CoVCS circumferential strain coefficient of variation, GCS global circumferential strain, CoVRS radial strain coefficient of variation, GRS global radial strain, KM Kaplan-Meier, CV cardiovascular.

• 

  Fig. 6. Strain biomarker KM hazard curves by tertiles for incident MI. In participants with reduced LVEF < 50% (panel A), strain CoV and global strain were not predictive of incident MI. Conversely, in participants with preserved LVEF ≥ 50% (Panel B), lower tertiles of GLS, GCS, GRS and greater tertiles of CoVLS were associated with greater risk of MI. LVEF left ventricular ejection fraction, CoVLS longitudinal strain coefficient of variation, GLS global longitudinal strain, CoVCS circumferential strain coefficient of variation, GCS global circumferential strain, CoVRS radial strain coefficient of variation, GRS global radial strain, KM Kaplan-Meier, MI myocardial infarction.

• 

  Fig. 7. Strain biomarker KM hazard curves by tertiles for incident HF. In participants with reduced LVEF < 50% (panel A),CoVCS and CoVRS stratified the risk of incident HF better than global strain. Conversely, in participants with preserved LVEF ≥ 50% (Panel B), the greatest tertile of CoVCS and CoVRS had comparable HF event numbers to the lowest tertile of GLS, GCS and GRS. LVEF left ventricular ejection fraction, CoVLS longitudinal strain coefficient of variation, GLS global longitudinal strain, CoVCS circumferential strain coefficient of variation, GCS global circumferential strain, CoVRS radial strain coefficient of variation, GRS global radial strain, KM Kaplan-Meier, HF heart failure

• 

  Fig. 8. Strain biomarker KM hazard curves by tertiles for incident arrhythmia. In participants with reduced LVEF < 50% (panel A), CoVCS and CoVRS stratified the risk of incident arrhythmia better than global strain. Conversely, in participants with preserved LVEF ≥ 50% (Panel B), GLS, GCS and GRS stratified the risk of incident arrhythmia better than their strain CoV counterparts. LVEF left ventricular ejection fraction, CoVLS longitudinal strain coefficient of variation, GLS global longitudinal strain, CoVCS circumferential strain coefficient of variation, GCS global circumferential strain, CoVRS radial strain coefficient of variation, GRS global radial strain, KM Kaplan-Meier

• 

  Fig. 9. Strain biomarker KM hazard curves by tertiles for death from any cause. In participants with reduced LVEF < 50% (panel A), CoVRS stratified the risk of death from any cause better than other strain markers. Conversely, in participants with preserved LVEF ≥ 50% (Panel B),GLS was the most sensitive marker for increased risk of all-cause death. LVEF left ventricular ejection fraction, CoVLS longitudinal strain coefficient of variation, GLS global longitudinal strain, CoVCS circumferential strain coefficient of variation, GCS global circumferential strain, CoVRS radial strain coefficient of variation, GRS global radial strain, KM Kaplan-Meier