Machine Learning Prediction of Cardiac Resynchronisation Therapy Response From Combination of Clinical and Model-Driven Data

Background: Up to 30–50% of chronic heart failure patients who underwent cardiac resynchronization therapy (CRT) do not respond to the treatment. Therefore, patient stratification for CRT and optimization of CRT device settings remain a challenge.Objective: The main goal of our study is to develop a predictive model of CRT outcome using a combination of clinical data recorded in patients before CRT and simulations of the response to biventricular (BiV) pacing in personalized computational models of the cardiac electrophysiology.Materials and Methods: Retrospective data from 57 patients who underwent CRT device implantation was utilized. Positive response to CRT was defined by a 10% increase in the left ventricular ejection fraction in a year after implantation. For each patient, an anatomical model of the heart and torso was reconstructed from MRI and CT images and tailored to ECG recorded in the participant. The models were used to compute ventricular activation time, ECG duration and electrical dyssynchrony indices during intrinsic rhythm and BiV pacing from the sites of implanted leads. For building a predictive model of CRT response, we used clinical data recorded before CRT device implantation together with model-derived biomarkers of ventricular excitation in the left bundle branch block mode of activation and under BiV stimulation. Several Machine Learning (ML) classifiers and feature selection algorithms were tested on the hybrid dataset, and the quality of predictors was assessed using the area under receiver operating curve (ROC AUC). The classifiers on the hybrid data were compared with ML models built on clinical data only.Results: The best ML classifier utilizing a hybrid set of clinical and model-driven data demonstrated ROC AUC of 0.82, an accuracy of 0.82, sensitivity of 0.85, and specificity of 0.78, improving quality over that of ML predictors built on clinical data from much larger datasets by more than 0.1. Distance from the LV pacing site to the post-infarction zone and ventricular activation characteristics under BiV pacing were shown as the most relevant model-driven features for CRT response classification.Conclusion: Our results suggest that combination of clinical and model-driven data increases the accuracy of classification models for CRT outcomes.

Better CRT Response in Patients Who Underwent Atrioventricular Node Ablation or Upgrade From Pacemaker: A Nomogram to Predict CRT Response

Background: Response rates for cardiac resynchronization therapy (CRT) in patients without intrinsic left bundle-branch block (LBBB) morphology are poor.Objective: We sought to develop a nomogram model to predict response to CRT in patients without intrinsic LBBB.Methods: We searched electronic health records for patients without intrinsic LBBB who underwent CRT at Mayo Clinic. Logistic regression and Cox proportional hazards regression analysis were performed for the odds of response to CRT and risk of death, respectively. Results were used to develop the nomogram model.Results: 761 patients without intrinsic LBBB were identified. Six months after CRT, 47.8% of patients demonstrated improvement of left ventricular ejection fraction by more than 5%. The 1-, 3-, and 5-year survival rates were 95.9, 82.4, and 66.70%, respectively. Patients with CRT upgrade from pacemaker [odds ratio (OR), 1.67 (95% CI, 1.05–2.66)] or atrioventricular node (AVN) ablation [OR, 1.69 (95% CI, 1.09–2.64)] had a greater odds of CRT response than those patients who had new implant, or who did not undergo AVN ablation. Patients with right bundle-branch block had a low response rate (39.2%). Patients undergoing AVN ablation had a lower mortality rate than those without ablation [hazard ratio, 0.65 (95% CI, 0.46–0.91)]. Eight clinical variables were automatically selected to build a nomogram model and predict CRT response. The model had an area under the receiver operating characteristic curve of 0.71 (95% CI, 0.63–0.78).Conclusions: Among patients without intrinsic LBBB undergoing CRT, upgrade from pacemaker and AVN ablation were favorable factors in achieving CRT response and better long-term outcomes.

Epicardial Adipose Tissue Measured From Computed Tomography Predicts Cardiac Resynchronization Therapy Response in Patients With Non-ischemic Systolic Heart Failure

Background: Epicardial adipose tissue (EAT) has been linked with the pathogenesis of heart failure (HF). Limited data have been reported about the clinical value of EAT for cardiac resynchronization therapy (CRT) in non-ischemic systolic HF. We aimed to explore the values of EAT measured from CT to predict the response to CRT in patients with non-ischemic systolic HF.Methods: Forty-one patients with CRT were consecutively recruited for our study. All patients received both gated resting Single Photon Emission CT (SPECT) myocardial perfusion imaging (MPI) and dual-source multi-detector row CT scans. EAT thickness was assessed on both the parasternal short and horizontal long-axis views. The area of EAT was calculated at the left main coronary artery level. Left ventricular systolic mechanical dyssynchrony (LVMD) was measured by phase standard deviation (PSD) and phase histogram bandwidth (PBW). The definition of CRT response was an improvement of 5% in left ventricular ejection fraction (LVEF) at 6 months after CRT implantation.Results: After 6 months of follow-up, 58.5% (24 of 41) of patients responded to CRT. A greater total perfusion deficit (TPD) was observed in the left ventricle, and a narrower QRS complex was observed in the nonresponse group than in the response group (p < 0.05). Meanwhile, the systolic PSD and systolic PBW were statistically greater in the CRT group with no response than in the response group (p < 0.05). Meanwhile, the baseline QRS duration, TPD, systolic PSD, systolic PBW, EAT thicknesses of the left ventricular (LV) apex, right atrioventricular (AV) groove, and left AV groove were all significantly related to the CRT response in the univariate logistic regression analysis. Furthermore, the QRS duration and EAT thicknesses of the right AV groove and left AV groove were independent predictors of CRT response in the multivariate logistic regression analysis.Conclusions: The EAT thickness of the left AV groove in patients with non-ischemic systolic HF is associated with the TPD of LV and LV systolic dyssynchrony. The EAT thickness of the AV groove has a good predictive value for the CRT response in patients with non-ischemic systolic HF.

Response to cardiac resynchronisation therapy in men and women: a secondary analysis of the SMART-AV randomised controlled trial

ObjectivesThere is a controversy about whether both sexes’ response to cardiac resynchronisation therapy (CRT) is similar. We aimed to assess a causal effect of sex on CRT response.DesignSecondary analysis of a randomised controlled trial (RCT) data. Doubly robust augmented-inverse-probability-weighted (AIPW) estimation of sex effect on CRT response.SettingThe SmartDelay Determined Atrioventricular (AV) Optimisation (SMART-AV) RCT.ParticipantsThe SMART-AV RCT enrolled New York Heart Association class III-IV patients with heart failure (HF) with left ventricular ejection fraction (LVEF) ≤35% despite optimal medical therapy and QRS duration ≥120 ms, in sinus rhythm. After exclusion of those with missing outcome or covariates, 741 participants (age 66±11 years; 33% female; 78% white; LVEF 28%±9%; 58% ischaemic cardiomyopathy; 75% left bundle branch block; left ventricular end-systolic volume index (LVESVI) 65±30 mL/m2) were included.InterventionsImplanted CRT defibrillator with randomly assigned AV delay as either (1) fixed at 120 ms, or (2) echocardiography-determined, or (3) SmartDelay algorithm-programmed.OutcomeA composite of freedom from death and HF hospitalisation and a >15% reduction in LVESVI at 6 month post-CRT was the endpoint.ResultsThe primary endpoint was met by 337 patients (45.5%); 134 were women (55.6% response) and 203 were men (40.6% response); p<0.0001. After conditioning for 33 covariates that included baseline demographic, clinical, ECG, echocardiographic and biomarker characteristics, known predictors of CRT response, logistic regression showed a higher probability for composite CRT response for women versus men (OR 1.79; 95% CI 1.08 to 2.98; p<0.0001), whereas AIPW estimation showed no difference in CRT response (average treatment effect 0.88; 95% CI 0.41 to 1.89; p=0.739). After removing colliders from the model, both logistic regression (OR 1.00; 95% CI 0.69 to 1.44) and AIPW (ATE 1.06; 95% CI 0.96 to 1.16) reported similar results.ConclusionsBoth sexes’ response to CRT is similar. Sex differences in HF substrate, treatment and comorbidities explain sex disparities in CRT outcomes.Trial registration numberClinicalTrials.gov Identifier; NCT00677014.

Right ventricular dysfunction is a predictor of non-response to cardiac resynchronization therapy

Introduction Cardiac resynchronization therapy (CRT) has been of great benefit to many heart failure (HF) patients with reduced ejection fraction (EF) and intraventricular conduction delay. However, approximately 30% of patients fail to respond to CRT. We investigated baseline characteristics that might influence response to CRT. Methods We retrospectively enrolled 227 patients undergoing CRT implantation between 2013 and 2020 according to the guidelines. 118 patients were included in our analysis, from whom all data were available. Clinical, electrocardiographic and echocardiographic parameters were evaluated at baseline and 6 months after CRT implantation. Response to CRT was defined as an increase in left ventricular ejection fraction (LVEF) &gt;10%. Right ventricular systolic dysfunction (RVSD) was defined as S' velocity &lt;9.5 cm/s or tricuspid anular plane systolic excursion (TAPSE) &lt;17 mm. Chronic kidney disease (CKD) was defined as GFR &lt;60 ml/min/1.73m2. Results 118 patients were included (mean age 69±11 years, 66.1% males, 39.8% ischemic etiology; 35,6% atrial fibrillation, baseline LVEF 27,6±6%). After 6 months of CRT, 65 patients (55.1%) were considered responders. Responders were more frequently female than non responders (43,1% vs 22,6, p=0.02). Atrial fibrillation and CKD were more prevalent in non responders (47,2% vs 26,2%, p=0.018; 62,3% vs 21,5%, p&lt;0.001, respectively). RVSD was present in 60,4% of non responders vs 16,9% of responders (p&lt;0.001). In responder group, the mean S' velocity was 10,9±2,1 cm/s vs 9,1±2,1 cm/s in non responder group, p&lt;0.001. The mean TAPSE was also higher in responder group (20,3±7,2 mm vs 16,5±4,4 mm, p=0.031). On multivariate analysis only RVSD (OR 7,754; 95% CI 2,968 – 20,282 p&lt;0.001] and CKD (OR 5,434; 95% CI 2,109 – 14,002; p&lt;0.001) were independently associated with non-response to CRT. Conclusion From a range of preoperative characteristics, multivariate analysis only identified RVSD and CKD as independent predictors of CRT response, with S' &lt;9,5 cm/s and TAPSE &lt;17 mm associated with non-response to CRT. This study highlights the importance of routine RV assessment in order to improve patient selection and optimize CRT response in heat failure patients. FUNDunding Acknowledgement Type of funding sources: None.

Machine Learning prediction of cardiac resynchronisation therapy response from combination of clinical and model-driven data

BackgroundUp to 30%-50% of chronic heart failure patients who underwent cardiac resynchronization therapy (CRT) do not respond to the treatment. Therefore, patient stratification for CRT and optimization of CRT device settings remain a challenge.ObjectiveThe main goal of our study is to develop a predictive model of CRT outcome using a combination of clinical data recorded in patients before CRT and simulations of the response to biventricular (BiV) pacing in personalized computational models of the cardiac electrophysiology.Materials and MethodsRetrospective data from 57 patients who underwent CRT device implantation was utilized. Positive response to CRT was defined by a 10% increase in the left ventricular ejection fraction in a year after implantation. For each patient, an anatomical model of the heart and torso was reconstructed from MRI and CT images and tailored to ECG recorded in the participant. The models were used to compute ventricular activation time, ECG duration and electrical dyssynchrony indices during intrinsic rhythm and BiV pacing from active poles of leads. For building a predictive model of CRT response, we used clinical data recorded before CRT device implantation together with model-derived biomarkers of ventricular excitation in the left bundle branch block mode of activation and under BiV stimulation. Several Machine Learning (ML) classifiers and feature selection algorithms were tested on the hybrid dataset, and the quality of predictors was assessed using the area under receiver operating curve (ROC AUC). The classifiers on the hybrid data were compared with ML models built on clinical data only.ResultsThe best ML classifier utilizing a hybrid set of clinical and model-driven data demonstrated ROC AUC of 0.82, an accuracy of 0.82, sensitivity of 0.85, and specificity of 0.78, improving quality over that of ML predictors built on clinical data from much larger datasets. Distance from the LV pacing site to the post-infarction zone and ventricular activation characteristics under BiV pacing were shown as the most relevant model-driven features for CRT response classification.ConclusionOur results suggest that combination of clinical and model-driven data increases the accuracy of classification models for CRT outcomes.

Five-year survival and biomarkers of sympatho-adrenal, neurohumoral, immune activation, fibrosis in patients with early and late superresponse to cardiac resynchronization therapy

Methods. 82 SR (80.5% men; mean age 60.4±9.3; 45 (54.9%) with CAD) were divided according to period of LVESV maximum decrease: Gr.1 (n=19)- <24months (14.0 [8.0; 21.0]), Gr.2 (n=63)->24 months (59[43.0; 84.0]). Dynamics of echocardiography, adrenaline (ADR) plasma levels, norepinephrine (NE), interleukins (IL) 1β, 6, 10, TNF-α, NT-proBNP, MMP-9, TIMP-1, 4, were examined. Five-year survival was estimated by Kaplan-Meier method. ROC analysis and logistic regression were applied to identify late CRT response factors.Results. Initially, groups didn’t differ by clinical and echocardiographic findings. At baseline, Gr.2 had larger ADR (p=0.049) and NE (p=0.061). Rate of change in NE was opposite in groups during CRT: ΔNE increased in Gr.1 and decreased in Gr.2. (p=0.015), which was associated with better reverse cardiac remodeling (lower LV end systolic diameter, LV end diastolic diameter, LVESV, LV end diastolic volume), decrease in activity of immune inflammation (decrease in levels of IL-1β, 6, 10, TNF-α) and fibrosis formation (decrease in TIMP- 1, enhancement of MMP-9/TIMP-1). Cut-off value of 2.55 ng/ml for NE complied with the highest sensitivity (80%), specificity (60%), AUC=0.693 (p=0.011) for predicting late CRT response. Proportion of patients with NE<2.55 ng/ml was 21.1% in Gr.1 and 59.7% in Gr.2, (p=0.003), mean follow-up period was 45.8±0.3 and 94.9±35.9 months (p<0.001), respectively All SR of Gr.2 were alive within 5 years, survival rate was 50% in Gr.1 (Log-Rank test<0.001). NE was associated with late CRT response (OR 8.0 (95%CI 1.5-42.8), p=0.015).Conclusion. Late CRT response was accompanied by increased life expectancy, better 5-year survival, associated with greater reverse cardiac remodeling, decreased fibrosis activity, immune, neurohumoral, sympathoadrenal activation. When NE level was less than 2.55 ng/ml, probability of late response increased 8-fold.