Prognostic Evaluation And Mechanism Study Of Heart Failure Device Therapy

Part I.Prognostic Evaluation of Heart Failure Device Therapy1.Association of Baseline Big Endothelin-1 Level with Long-term Prognosis among Cardiac Resynchronization Therapy RecipientsObjectives:To assess the effect of big endothelin-1(ET-1)on long-term clinical outcomes in patients receiving cardiac resynchronization therapy(CRT).Methods:A cohort of 367 consecutive patients who received CRT between January 2010 and December 2015 were enrolled,and categorized into three groups according to baseline big ET-1 tertiles:big ET-1≤0.34 pmol/L(N=119),big ET-1 between 0.34-0.56 pmol/L(N=127)and big ET-1>0.56 pmol/L(N=121).The endpoints defined as mortality rate(all-cause)and heart transplantation,which were analyzed using Kaplan-Meier and log-rank test curves.Cox proportional-hazards regression models was used to explore independent prognostic factors.Results:Over a median follow-up of 21 months,48(13.08%)patients died,6(1.63%)underwent heart transplantation and 100(27.25%)had heart failure hospitalization(HFH).We found a significant difference in event free survival between the three groups,with high levels of big ET-1 correlating with worse survival(Log-rank test,P<0.001).After adjusting for multiple factors in the multivariate model,big ET-1>0.56pmol/L was an independent predictor for primary endpoint event[hazard ratio(HR):2.005,95%confidence interval(CI)1.045-6.2621,P=0.040]and HFH(HR=2.126,95%CI 1.182-3.827,P=0.012).Conclusions:Baseline big ET-1>0.56 pmol/L was independently associated with higher all-cause mortality and HFH among CRT recipients,and therefore can be added to the marker panel used for stratifying high risk CRT patients for priority treatment.2.EAARN,VALID-CRT and ScREEN Risk Score to Predict Long-term Outcomes for Patients Receiving Cardiac Resynchronization Therapy in an Chinese PopulationObjectives:To validate externally and recalibrate 3 European risk scores for all-cause mortality and transplantation in patients receiving cardiac resynchronization therapy(CRT)in an Asian population.Methods:Data were collected at our institution between January 2010 to December 2017.The primary endpoints were all-cause mortality and heart transplantation.Kaplan-Meier analyses for each group and score were analyzed respectively.Discrimination and calibration of EAARN score system were evaluated though c-statistics and Hosmer-Lemeshow(H-L)goodness-of-fit test.Results:Of the 506 patients who were followed for 2 years,104 reached the primary endpoint The Kaplan-Meier event-free survival analysis,stratified according to the three scores,yielded significant results(log-rank test,all P<0.05),with a good fit between the predicted and observed event rates(H-L goodness-of-fit test,all P>0.05).The ScREEN score[AUC:0.643(0.582-0.704),c-statastics:0.612(0.606-0.618)]yielded the best discriminatory power for the primary endpoints compared to the VALID-CRT[AUC:0.596(0.0.539-0.654),c-statistics:0.574(0.519-0.629)]and EAARN[AUC:0.626(0.565-0.687),c-statistics:0.582(0.521-0.643)]scores.Conclusions:ScREEN was the best predictor of all-cause mortality and heart transplantation.Risk scores based on different populations should be selected cautiously.3.Machine learning with Biomarkers Improves the Prediction of Long-term Outcomes in Patients underwent Cardiac Resynchronization TherapyObjectives:We aimed to improve the prediction of long-term all-cause mortality and heart transplantation in HF patients and optimize risk stratification by developing a machine-learning(ML)model using pre-implantation data and advanced laboratory biomarkers.Methods:We retrospectively selected 506 consecutive HF patients who were treated at our institution between January 2010 and December 2017.More than 25,000 parameters,including demographic data,and data on electrocardiography parameters,index of HF,echocardiography,and advanced laboratory biomarkers were collected before implantation and used to train six ML models.Model performance was evaluated using ten-fold cross-validation.Primary endpoint was all-cause mortality.Compared with the models with basic inputs,the areas under the curves were improved after adding 16 advanced laboratory biomarker features.The random forest model(RF-CRT)performed the best on predicting all-cause mortality(ROC:0.75).The stratification of RF-CRT risk in quartiles enabled significant discrimination of the risk of primary endpoint and HF hospitalization(log-rank,P<0.001).Results:Compared with the models with basic inputs,the areas under the curves were improved after adding 16 advanced laboratory biomarker features(0.71±0.08 vs 0.75±0.09).The random forest model(RF-CRT)performed the best on predicting all-cause mortality(ROC:0.75±0.09).The stratification of RF-CRT risk in quartiles enabled significant discrimination of the risk of primary endpoint and HF hospitalization(log-rank,P<0.001)Conclusions:ML models improved the prediction of long-term all-cause mortality and heart transplantation in HF patients before CRT,and risk stratification of CRT recipients.Part Ⅱ.Underlying Mechanism of Heart Failure Device Therapy1.Comprehensive plasma metabolites profiling reveals phosphatidylcholine species as potential predictors for CRT responseObjectives:The purpose of this study was to identify plasma metabolite fingerprint in HF patients,and develop a prediction tool based on differential metabolites for the response of CRT.Methods:We prospectively recruited 32 healthy individuals and 42 consecutive HF patients who underwent CRT between January 2018 and January 2019.Peripheral venous blood samples,clinical,echocardiographic evaluation were collected for each patient before CRT implantation.Liquid chromatography-mass spectrometry(LC-MS/MS)was used to analyze metabolites.After 6 months of follow-up,patients were categorized as CRT responders and non-responders based on comparative echocardiographic measurements.Results:Compared to healthy individuals,patients with heart failure had a distinct metabolomic profiles,including free fatty acid,carnitine,amino acid and lipid.A distinct baseline metabolomic profile was observed between CRT responders and non-responders.For differential diagnosis,a panel of 4 phosphatidylcholine metabolites provided areas under the curve of 0.906 for CRT response.The optimal cutoff off was 0.275(P<0.001),with 83.3%sensitivity and 90.0%specificity.Conclusions:Metabolomics is useful for identifying differential metabolites in HF patients,and a panel of 4 phosphatidylcholine species metabolites may be an applicable tool for risk stratification among HF patients treated with CRT.2.The Effect of Cardiac Contractility Modulation on Metabolomics in Canine Model of Heart Failure.Objectives:To observe the effect of Cardiac contractility modulation(CCM)on metabolomics in canine model of hyper-pacing heart failureMethods:Seventeen healthy beagle dogs were selected and divided into sham operation group,heart failure control group,and CCM treatment group.Preoperative echocardiography was performed to measure the size of the heart chambers,the wall thickness,left ventricular ejection fraction(LVEF)and heart rate detected by electrocardiogram.The heart failure control group and CCM treatment group were implanted with pacing leads via the right external jugular vein and placed in a hyper-pacing pacemaker.The heart failure model was prepared for 6 weeks of fast paced pacing.In the sham group,only the external jugular vein was isolated and no pacing lead was placed.After 6 weeks of hyper pacing,the pacemaker in the CCM treatment group was replaced with CCM,while the pacemaker in heart failure control group was turned off.One week later the echocardiogram and electrocardiogram were reexamined.Pathological analysis was performed by anatomy.The pathological changes of left ventricle were observed by hematoxylin-eosin and Masson’s staining.We applied a comprehensive metabolomics platform to myocardial samples to assess the metabolomics profiles in the three groups.Results:All 17 dogs successfully completed the entire experiment.After 6 weeks of hyper-pacing,cardiac lumen enlargement in the heart failure control group were observed.The modeling procedure of the heart failure model was successful.After 1 week of echocardiogram examination,the increase of LVEF in the CCM group was significantly higher than that in the control group(CCM vs HF 51.8±8.16%vs 39.2±4.53%,P=0.002)Pathology suggests that CCM treatment has some improvement in left ventricular remodeling caused by heart failure.In addition,we also found that and ICM exhibited different metabolomics profiling,and the disease specific differential metabolites cluster the samples into diseases.Pathway analysis revealed that the linoleic acid metabolism,sphingolipid metabolism and arachidonic acid metabolism were ICM-specific.Meanwhile,D-Glutamine and D-glutamate metabolism DCM-specific,highlighting the role of metabolic dysregulation related to different disease pathogenesis.Conclusions:CCM can improve heart function,hemodynamic parameters,and delay left ventricular remodeling caused by heart failure in dogs with hyper-pacing model of heart failure.