Novel cardiovascular magnetic resonance strain heterogeneity phenotypes predict cardiovascular events: A prospective UK Biobank study.
Authors
Affiliations (7)
Affiliations (7)
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK; James Cook University Hospital, Marton Road, Middlesbrough TS4 3BW. Electronic address: [email protected].
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK; Newham University Hospital, Barts Health NHS Trust, Glen Road, Plaistow, London E13 8SL, UK; Barts Heart Centre, St Bartholomew's Hospital, Barts Health NHS Trust, West Smithfield, EC1A 7BE, London, UK.
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK; Barts Heart Centre, St Bartholomew's Hospital, Barts Health NHS Trust, West Smithfield, EC1A 7BE, London, UK; Center for Inherited Cardiovascular Diseases, WellSpan Health, York, PA, USA; Department of Cardiovascular Medicine, Mayo Clinic, Rochester, MN, USA.
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK; Barts Heart Centre, St Bartholomew's Hospital, Barts Health NHS Trust, West Smithfield, EC1A 7BE, London, UK.
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK.
- DC Veterans Affairs Medical Center, Washington DC.
- William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary University London, Charterhouse Square, London, EC1M 6BQ, UK; Barts Heart Centre, St Bartholomew's Hospital, Barts Health NHS Trust, West Smithfield, EC1A 7BE, London, UK. Electronic address: [email protected].
Abstract
Global longitudinal (GLS), circumferential (GCS) and radial (GRS) strains may be insufficiently sensitive to early regional pathological cardiac remodeling on cardiovascular magnetic resonance imaging (CMR). Corresponding strain coefficients of variation, CoV<sub>LS</sub>, CoV<sub>CS</sub> and CoV<sub>RS</sub>, may facilitate improved prognostication by quantifying contractile heterogeneity. To compare CoV<sub>LS</sub>, CoV<sub>CS</sub>, CoV<sub>RS</sub>, to GLS, GCS, GRS, to predict incident cardiovascular (CV) events at population level, respectively. CMR feature tracking-derived strain biomarkers from 60,746 UK Biobank participants were calculated. Kaplan-Meier survival analysis and Cox proportional hazards regression analyzed unadjusted and adjusted associations between strain biomarkers and CV outcomes, respectively. Covariables included age, sex, CV risk factors, left ventricular mass, end-diastolic volume, ejection fraction (LVEF) and pre-existent regional strain abnormalities. Logrank test for trend, hazard ratios (HR) and Uno's C-Index compared relative performances of CoV and global strain. Over a median follow-up 5.1 years, higher CoV<sub>CS</sub>, CoV<sub>RS</sub> and lower GLS predicted greater risk of all-cause death (HR 1.10 [1.02-1.19]; HR 1.08 [1.01-1.16]; HR 0.84 [0.77-0.93], respectively) and a composite CV endpoint of myocardial infarction (MI), heart failure (HF) and arrhythmia (HR 1.10 [1.04-1.16]; HR 1.06 [1.01-1.12]; HR 0.77 [0.71-0.82], respectively). Model discrimination of HF and arrhythmia were significantly improved by CoV<sub>CS</sub> (△C-index 0.004 [P<0.001], 0.002 [P=0.001], respectively) and CoV<sub>RS</sub> (△C-index 0.004 [P<0.001], 0.001 [P=0.004], respectively). When LVEF ≥50% and ≥3 of age >65 years, obesity, hypertension, diabetes, atrial fibrillation and chronic kidney disease, CoV<sub>CS</sub> and CoV<sub>RS</sub> were more predictive of incident HF than GLS (HR 1.37 [1.14-1.65]; HR 1.35 [1.14-1.60]; HR 0.68 [0.53-0.86], respectively). When LVEF <50%, CoV<sub>CS</sub> and CoV<sub>RS</sub> were superior to GLS in predicting the composite CV endpoint, HF and arrhythmia (logrank test for trend, P<0.001 for all with CoV<sub>CS</sub> and CoV<sub>RS</sub>vs P=0.022, P=0.15, P=0.030 for GLS respectively). Heterogeneity biomarkers are sensitive to early pathological signals by measuring disease regionality. CoV<sub>CS</sub> and CoV<sub>RS</sub> are significant, consistent, additive and sometimes superior predictors of HF, arrhythmia and all-cause death than established risk markers, particularly in cohorts with multiple co-morbidities or LVEF <50%. Expansion of machine learning-guided image analysis makes strain CoV imminently translatable into routine clinical practice.