Unsupervised machine learning generated clusters of left ventricular strain curves identifies patients in risk of heart failure and cardiovascular death following acute myocardial infarction

Background Today myocardial deformation, also known as strain, is assessed by the global longitudinal strain (GLS) which only provides information about the maximal deformation during systole. Hence, a lot of information obtained from different patterns of deformation curves might be undiscovered. Unsupervised Machine leaning (uML) is capable of identifying similar patterns of deformation curves. Identifying different phenotypical patterns from myocardial deformation curves might provide insights into the pathophysiological development of cardiac disease and entail useful prognostic information. Purpose To investigate whether uML can group specific patterns of myocardial deformation curves which provide prognostic information on heart failure and/or cardiovascular death (HF/CVD) following ST-segment elevation myocardial infarction (STEMI). Methods A total of 319 STEMI patients had an echocardiogram performed at median 2 days after primary percutaneous coronary intervention (pPCI). Speckle tracking echocardiography analysis divided the left ventricle into 18 segments. Standardisation of the cardiac cycle was done using linear interpolation and complete strain data (mean of all segments) as function of time throughout the cardiac cycle was used as input for the uML algorithm. Clusters were identified using a K-means cluster analysis algorithm. Primary endpoint was the composite of heart failure (HF) and/or cardiovascular death (CVD). Median follow-up time was 1423 days (IQR: 91; 1660). Results Mean age was 62 years, 75% were male and 130 (41%) suffered incident HF/CVD during follow-up. The uML algorithm grouped patients into three clusters containing 97, 104, and 118 patients respectively. GLS curves of the three clusters are illustrated in the Figure 1. Incidence of HF/CVD increased significantly from cluster 1 through 3 (24% vs. 39% vs. 60%, P<0.001). In multivariable Cox regressions adjusting for the variables in the score risk chart model all three clusters were significantly associated with future HF/CVD (Figure 1). Cluster models provided significant incremental prognostic information when comparing C-statistics (0.64 vs. 0.62, p=0.029) Conclusion Unsupervised Machine Learning clusters of left ventricular deformation curves identifies patients in risk of HF/CVD following STEMI treated with pPCI, and provides incremental prognostic information to the score risk chart model. Figure 1. GLS curves of the three clusters Funding Acknowledgement Type of funding source: None

The prognostic value of left ventricular mechanical dyssynchrony in predicting incident atrial fibrillation and ischemic stroke in the general population

Background Left ventricular mechanical dyssynchrony (LVMD) is a predictor of many cardiovascular outcomes including ventricular arrhythmias. However, the prognostic value of LVMD in predicting incident atrial fibrillation (AF) in participants from the general population is currently unknown. Purpose The aim of this study was to investigate if LVMD can be used to predict AF and ischemic stroke in the general population. Methods A total of 1282 participants (mean age 57±16 years, 42% male) from the general population underwent a health examination including two-dimensional speckle tracking echocardiography. LVMD was calculated as the standard deviation of the regional time-to-peak strain from the three apical views. The primary endpoint was incident AF at follow-up. All participants with known AF and prior stroke at baseline were excluded (n=84). The secondary endpoint consisted of the composite of AF and ischemic stroke. Results During a median follow-up of 16 years, 148 participants (12%) were diagnosed with incident AF and 88 (7%) experienced an ischemic stroke, resulting in 236 (19%) experiencing the composite outcome. The risk of AF increased incrementally with increasing tertile of LVMD, being approximately 2-fold higher in the 3rd tertile as compared to the 1st tertile (HR 1.79; 95% CI (1.22–2.63), p=0.003; figure). LVMD was a univariable predictor of AF with 7% increased risk per 10ms increase in LVMD (per 10ms: HR 1.07; 95% CI (1.03–1.12), p<0.001). The association remained significant even after multivariable adjustment for age, sex, body mass index, hypertension, diabetes, previous ischemic heart disease, systolic blood pressure, diastolic blood pressure, heart rate, smoking, plasma proBNP, left ventricular ejection fraction <50%, global longitudinal strain, left atrial volume index (LAVI) and E/e' (per 10ms increase: HR 1.06; 95% CI (1.01–1.12), p=0.018). LVMD was also a univariable predictor of the composite outcome of AF and ischemic stroke (per 10ms increase: HR 1.07; 95% CI (1.04–1.11), p<0.001). After multivariable adjustment for the same clinical and echocardiographic parameters, LVMD remained an independent predictor of the composite outcome (per 10ms: HR 1.07; 95% CI (1.03–1.11), p=0.001). Additionally, LVMD provided incremental prognostic information with regard to predicting AF as assessed by a significant increase in the net reclassification improvement (NRI) index beyond the CHARGE-AF score (continuous NRI, 0.300; 95% CI, 0.022–0.503). Furthermore, LVMD provided additional incremental prognostic information, when added to both the CHARGE-AF score and the LAVI (continuous NRI, 0.269; 95% CI, 0.004–0.499). Conclusion In a low risk general population, LVMD provides novel prognostic information on the long-term risk of AF and ischemic stroke. In addition, LVMD provides incremental prognostic information beyond the CHARGE-AF score and LAVI in predicting AF in the general population. Funding Acknowledgement Type of funding source: None

Incremental prognostic value of tricuspid annular dilatation over the STS score

Background Transcatheter aortic valve replacement (TAVR) is the treatment of choice in most patients with severe aortic stenosis. The Society of Thoracic Surgeons (STS) score is a well established risk score to estimate morbidity, mortality and procedural risk of patients undergoing TAVR. However, tricuspid annular Dilatation (TAD), which is an increasingly recognized pathology associated with increased mortality, is not implemented in the STS Score. Purpose The purpose of this analysis was to investigate the incremental prognoctic value of TAD over the STS score. Methods Maximal septo-lateral diameter of the tricuspid annulus was measured in 923 patients on 3-dimensional MDCT datasets. A cut-off of 23 mm/m2 body-surface area was revealed by receiver-operating curve statistics and used to define TAD. Incremental prognostic Information was tested with c-index statistics and continuous net reclassification improvement (NRI). Patients were followed for 2 years and all-cause mortality was defined as study endpoint. Results Of 923 patients included in this analyis, TAD was found in 370 patients (40%). Patients with TAD had a significantly higher mortality (hazard ratio 2.18 with 95% CI 1.71 and 2.78, p<0.001). The mean STS score in the investigated patient cohort was 5.6±5.0. TAD provided incremental prognostic Information over the STS score when assessed with c-index statistics (rise from 0.63 to 0.66, p<0.01) or continuous NRI (0.209 with 95% CI 0.127 and 0.292, p<0.001). Estimated survival rates at 2 years were 88.2% (95% CI 84.5 and 92.1) in patients with a low STS score (<4) and no TAD and 57.5% (95% CI 51.1 and 64.7) in patients with a high STS score (>4) and TAD. Estimated survival rates in patients with a low STS score and TAD and patients with a high STS score and no TAD were similiar (75.8% with 95% CI 68.9 and 83.5 and 74.8% with 95% CI 69.2 and 80.7, respectively). Kaplan-Meier curves are shown in Figure 1. Conclusion TAD is a common entity in patients undergoing TAVR for severe aortic stenosis. It is associated with significantly higher mortality and provides incremental prognostic Information over the STS score. Funding Acknowledgement Type of funding source: None

Incremental prognostic value of global myocardial work over ejection fraction and global longitudinal strain in patients with heart failure and reduced ejection fraction

Aims Left ventricular (LV) ejection fraction (EF) and global longitudinal strain (GLS) help identify heart failure (HF) patients who are at risk for adverse outcomes. This study aimed to determine whether global myocardial work (GMW), derived from non-invasive LV pressure-strain loops, can provide incremental prognostic information over EF and GLS in patients with HF and reduced EF (HFrEF). Methods and results We retrospectively analysed 508 patients (age 62.9 ± 15.8 years, 29.1% female) with LVEF ≤40%. The study endpoint was a composite of all-cause death and HF hospitalization. The incremental value of GMW over clinical and echocardiographic variables including EF and GLS for the association with the composite endpoint was assessed using Cox regression analyses. Over a 1-year follow-up, 183 patients reached the endpoint. Baseline variables associated with the endpoint were age, haemoglobin, LV end-systolic volume, New York Heart Association Class III or IV, E/e′ ratio, pulmonary artery systolic pressure, EF, and GLS. Cox regression analysis revealed that GMW [hazard ratio (HR) 1.15, 95% confidence interval (CI) 1.05–1.25, per 100-mmHg% decrease] added incremental prognostic value over these variables. Both EF and GLS were not independent variables when GMW was included in the model. Patients with GMW <750 mmHg% were associated with a significantly higher risk of all-cause death and HF hospitalization (HR 3.33, 95% CI 2.31–4.80) than patients with GMW ≥750 mmHg%. Conclusion In patients with HFrEF, GMW provides incremental prognostic information over EF and GLS regarding risk of all-cause death and HF hospitalization.

P2442Myocardial performance index predicts mortality in people with obstructive lung function from the general population

Background Forced expiratory volume in one second (FEV1) is a significant predictor of mortality in patients with obstructive lung function (OL). Whether echocardiography can be used to identify patients at high risk, and whether it provides incremental prognostic information on mortality in patients with OL, remains unknown. Methods In a large, low-risk general population study, 1873 participants underwent a health examination with spirometry and echocardiography, including tissue Doppler imaging (TDI). The myocardial performance index (MPI) was calculated as the sum of the isovolumic contraction time (IVCT) and the isovolumic relaxation time (IVRT) divided by the left ventricle ejection time (LVET). Spirometry included measurements of (FEV1) and the forced vital capacity (FVC). OL was defined as FEV1/FVC <0.70. The primary endpoint was all-cause mortality. Results The mean age was 59±16 years, 57% were women, 43% had hypertension, 11% had diabetes, and 6% had ischemic heart disease. Of the 1873 included participants, 288 (15%) were classified as having OL at baseline. During follow up (median 13.7 years (IQR 13.2–16.2)), 584 (31%) persons died, hereof 178 (62%) in the subgroup of participants with OL and 406 (26%) in the subgroup of participants with normal lung function. OL was associated with presence of left ventricular hypertrophy (higher left ventricular mass index), impaired diastolic function (lower E, higher A, lower E/A ratio, longer deceleration time, lower e' and higher E/e'), lower global longitudinal strain, and higher MPI. In unadjusted analysis, higher MPI was associated with all-cause mortality for participants with OL (HR=1.18 (1.11–1.26), p<0.001, per 0.1 increase) and for participants with normal lung function (HR=1.42 (1.34–1.50), p<0.001, per 0.1 increase). The predictive value of MPI was significantly modified by the presence of obstructive lung function (p<0.001). After multivariable adjustment for age, sex, FEV1/FVC, heart rate, systolic blood pressure, smoking status, body mass index (BMI), hypertension, diabetes, ischemic heart disease, ischemic stroke and heart failure at baseline, MPI remained an independent predictor of all-cause mortality (HR=1.19 (1.06–1.34), p=0.004, per 0.1 increase) for participants with OL but not for participants with normal lung function (HR=1.02 (0.94–1.11), p=0.598, per 0.1 increase). When adding the MPI to the updated Age, Dyspnea and Obstruction (ADO) index, MPI provided incremental prognostic information beyond the updated ADO index, as determined from a significant increase in the Harrell's C-statistics (0.785 to 0.792, p=0.003). Conclusion Presence of OL is associated with subtle impairment of left ventricular systolic function, impaired left ventricular diastolic function, and higher MPI. MPI is an independent predictor of mortality in people with OL and provides incremental prognostic information regarding all-cause mortality in this population. Acknowledgement/Funding Herlev & Gentofte University Hospital PhD fund

2138Usefulness of left atrial strain for predicting incident atrial fibrillation and ischemic stroke in the general population

Background Left atrial (LA) enlargement is an established independent predictor of incident atrial fibrillation (AF). However, the prognostic value of left atrial peak reservoir strain (LA RS) in predicting incident AF in participants from the general population is currently unknown. It is our hypothesis that decreased LA RS can reveal early atrial dysfunction. Purpose The aim of this study was to investigate if LA RS can be used to predict AF and ischemic stroke in the general population. Methods A total of 405 participants (mean age 56±16 years, 41% male) from the general population underwent a health examination including two-dimensional speckle tracking echocardiography of the LA. LA RS was calculated as the average from the three apical views. The primary endpoint was incident AF at follow-up. All participants with known AF and prior stroke at baseline were excluded (n=54). The secondary endpoint consisted of the composite of AF and ischemic stroke. Results During a median follow-up of 16 years (interquartile range, 13.6–16.2 years), 36 participants (9%) were diagnosed with incident AF and 30 (7%) experienced an ischemic stroke, resulting in 66 (16%) experiencing the composite outcome. The risk of AF increased incrementally with decreasing tertile of LA RS, being approximately 10-fold higher in the 1st tertile as compared to the 3rd tertile (HR 9.82; 95% CI (2.95–32.63), p<0.001; figure). LA RS was a univariable predictor of AF with 41% increased risk per 5% decrease in LA RS (per 5% decrease: HR 1.41; 95% CI (1.18–1.67), p<0.001). However, the prognostic value of LA RS was modified by age (p=0.002 for interaction). After adjusting for clinical and echocardiographic parameters the LA RS predicted AF in participants aged <65 years (per 5% decrease: HR 1.86; 95% CI (1.20–2.90), p=0.006). In contrast, LA RS did not predict AF in participants aged >65 years (per 5% decrease: HR 0.95; 95% CI (0.73–1.23), p=0.69). LA RS was also a univariable predictor of the composite outcome of AF and ischemic stroke (per 5% decrease: HR 1.29; 95% CI (1.14–1.46), p<0.001). After multivariable adjustment the LA RS predicted AF and ischemic stroke in participants aged <65 years (per 5% decrease: HR 1.33; 95% CI (1.03–1.72), p=0.030). Furthermore, LA RS provided incremental prognostic information over the left atrial volume index (LAVI) with regard to predicting AF (Harrell's C-statistics 0.69 vs. 0.75, p=0.044) and the composite of AF and ischemic stroke (Harrell's C-statistics 0.59 vs. 0.66, p=0.027) in participants from the general population. Conclusion In a low risk general population, the LA RS provides novel prognostic information on the long-term risk of AF and ischemic stroke, especially in participants aged <65 years. In addition, LA RS provides incremental prognostic information over the LAVI in predicting AF and ischemic stroke in the general population.

Prognostic value of ratio of transmitral early filling velocity to early diastolic strain rate in patients with Type 2 diabetes

Aims The ratio of early mitral inflow velocity to global diastolic strain rate (E/e′sr) has recently emerged as a novel measure of left ventricular filling pressure. E/e′sr has in previous studies demonstrated to have good prognostic value in various patient populations. The aim of this study is to investigate the prognostic value of E/e′sr in a large cohort of patients with Type 2 diabetes in relation to cardiovascular morbidity and mortality. Methods and results In this prospective study, 848 Type 2 diabetic patients (mean age 63.6 ± 10.3 years, 64.7% male) underwent comprehensive echocardiographic examination including 2D speckle tracking in which E/e′sr along with novel and conventional echocardiographic variables were obtained. During follow-up (median: 4.8 years, interquartile range: 4.0–5.3), 122 (14.1%) met the composite outcome of cardiovascular disease, hospitalization, and mortality. Both E/e′sr and E/e′ were significantly associated with the outcome [E/e′sr: hazard ratio (HR) 1.07, 95% confidence interval (CI): 1.05–1.10; P < 0.001, per 0.10 m increase] and (E/e′: HR 1.07, 95% CI: 1.05–1.10; P = 0.001, per 1 unit increase). E/e′sr remained an independent predictor after multivariable adjustment for demographical, clinical, and echocardiographic parameters (HR 1.06, 95% CI: 1.01–1.12; P = 0.032, per 10 cm increase). The same was true for E/e′ (HR 1.09, 95% CI: 1.04–1.14; P < 0.001, per 1 unit increase). Additionally, E/e′sr provided incremental prognostic information beyond the UK ‘Prospective Diabetes Study risk engine’ 0.72 (0.68–0.77) vs. 0.74 (0.70–79), P = 0.040. Conclusion In patients with Type 2 diabetes, E/e′sr provides independent and incremental prognostic information regarding cardiovascular morbidity and mortality.