Correlation Of Magnetic Resonance Characteristics With Genes And Prognosis In Hypertrophic Cardiomyopathy

Objectives Genetic testing is recommended in patients with hypertrophic cardiomyopathy(HCM)for family screening,however,the yield of causative mutations is viable,and it is time-consuming with certain economic burden.Thus it has not been widely used in clinical practice.Cardiac magnetic resonance(CMR)is widely used in the clinical management of HCM for evaluating cardiac structure and motion,myocardial perfusion,tissue characterization.The application of radiomics in medical images provides a new insight for the diagnosis and risk stratification in various diseases.Recent years,deep learning,a subtype of machine learning which uses complex artificial neural networks to directly process raw data,has always been at the forefront of radiomics.The combinations of deep learning method and multiple CMR sequences have been investigated for diagnosis and risk stratification in various cardiovascular disease.To improve the cost-effectiveness of genetic testing,we aim to apply a deep learning algorithm on non-enhanced cine images to select patients with a high probability of positive HCM genotypes.Methods We consecutively recruited 198 HCM patients(48%men,age 47±13 years)including an training set(147 cases from Department of CMR)and an test set(51 cases from Department of Cardiology).All patients underwent CMR examination,HCM genetic testing[including eight major HCM pathogenic genes:myosin binding protein C(MYBPC3),?-myosin heavy chain(MYH7),essential and regulatory myosin light chain(MYL2 MYL3),troponin T(TNNT2),troponin I(TNNI3),tropomyosin(TPM1),and ?-cardiac actin(ACTC)]and assessment of established genotype scores(Mayo Clinic score I,Mayo Clinic score ?,and Toronto score).LV myocardium were manually drawn by a ITK-SNAP 3.6.0 software on the end-systolic and end-diastolic phases of 4-chamber-view cine images,and a deep learning(DL)method was developed to extract information from cine images to predict positive HCM genotype.The predictive performance of using DL method and established HCM genotype scores in identifying positive HCM genotypes were compared using receiver-operating characteristic(ROC)analysis.In order to verify the accuracy of different methods,we performed a 10-fold cross-validation in the training set.The incremental value of DL method was assessed by using the integrated discrimination improvement(IDI)and net reclassification improvement(NRI)Results Ninety-eight(49.49%)of all the 198 HCM patients were genetic positive,and 76.53%of the pathogenic mutations were located in MYBPC3 and MYH7.The proportion of hypertension,hyperlipidemia and family history of HCM in the training group were significantly higher than those in the test group,while there were no significant differences in other clinical baseline data between the two groups.The area under curve(AUC)of different methods for identifying positive HCM genotype were:the Mayo Clinic score I(0.63±0.04,0.64),Mayo Clinic score ?(0.67±0.04,0.70),Toronto score(0.68±0.04,0.74),DL method(0.81±0.01,0.80)for the 10-fold cross-validation on the training and test set,respectively.The combination of DL method and Toronto score showed the highest predictive performance(AUC=0.84)and had incremental value compared with the established genotype scores alone(all p<0.05).Conclusions Utilizing the DL method based on non-enhanced cine CMR images has incremental value to highly select HCM patients with positive genotype.In addition,the combination of DL method and Toronto score had the highest predictive performance,which could help improve the cost-effectiveness of HCM genetic testing.Objectives Hypertrophic cardiomyopathy(HCM)is an inherited cardiovascular disease in which about half of the patients carry mutations in genes encoding sarcomeric proteins.However,clear-cut correlations have not been established between phenotype and genotype in HCM patients.Myocardial fibrosis is a characteristic histological finding of HCM.Late gadolinium enhancement(LGE)technology by cardiac magnetic resonance(CMR)can be used to noninvasively evaluate myocardial fibrosis,and LGE is also reported as an risk factor for cardiovascular events including sudden death,arrhythmia and heart failure in HCM patients.This study used CMR parameters with the aim to 1)investigate the correlation between genotype and CMR parameters HCM patients;2)investigate the differences in CMR parameters and prognosis in HCM patients with different genotypes;3)investigate the role of genotype and CMR parameters in HCM risk stratification.Methods This study recruited 179 HCM patients who received CMR examinations at our hospital from 2011 to 2013.All patients underwent HCM genetic testing(including eight major genes encoding sarcomere proteins:MYBPC3,MYH7,MYL2,MYL3,TNNT2,TNNI3,TPM1,ACTC),clinical(including gender,age,family history,symptoms,hypertension,hyperlipidemia,diabetes,etc.)and CMR data(including left ventricular functional parameters and mass,shape of interventricular septum,LGE extent,etc.).All patients were followed up by regular telephone or by reviewing inpatient/outpatient records.Endpoint events were defined as(1)malignant ventricular arrhythmia events:sudden cardiac death,persistent ventricular tachycardia and ventricular fibrillation,implantable cardioverter defibrillator(ICD)discharge;(2)heart failure events(heart functional progression,heart failure hospitalization or death,heart transplantation);(3)stroke.Independent sample t-test or nonparametric test,chi-square test or Fisher's exact test were used for comparisons of clinical and CMR parameter in patients with positive and negative genotype,and in patients with and without events,Kaplan-Meier curves were used to compare the prognosis of patients with positive and negative genotype,and the patients with different pathogenic genes.Cox regression analysis was used to determine the risk factors of cardiovascular events and the corresponding coefficient hazard ratio(HR).Results Eighty-eight patients had positive genotype,accounting for 49.2%of all the HCM patients.After a median follow-up for 5.81(4.35-6.35)years,33 patients experienced endpoint events,including 7 cases of sudden cardiac death,1 case of ICD discharge,18 cases of heart failure rehospitalization or cardiac function progression,2 cases of heart failure death,and 5 cases of stroke.The left ventricular maximal wall thickness(LVMWT)were greater in patients with positive genotype than that of the patients with negative genotype(24.00 ± 5.77mm vs 21.60±5.69mm,p=0.002).Patients with positive genotype had significantly higher proportion of LGE presence(92.0%vs 73.6%,p=0.001)and the LGE extent[11.01(5.56-17.02)%vs 5.20(0.00-11.09)%,p<0.001]than that of the patients with negative genotype.The gene positive rate was 34.5%,55.4%,60.0%in different LVMWT subgroups:<20,[20-30),?30mm,34.5%,55.4%,60.0%in the group,<10%,[10%-20%],and was 40.2%,48.0%,59.3%in different LGE extent subgroups:<10%,[10%-20%),?20%,respectively.Among all the four ventricular septum shapes,the gene positive rate was the lowest(22.7%)in the apical type and the highest in the reversed curve type(58.0%).The prognosis of patients with positive genotype was worse than that of patients with negative genotype(log rank p=0.023),while the CMR parameters and prognosis were not significantly different in the patients with MYBPC3 and MYH7 mutations.In univariate Cox regression analysis,male gender,syncope,LGE extent(per 10%)and positive genotype were associated with cardiovascular events in HCM,however,in multivariate Cox regression analysis,male gender[HR:2.16(1.07 4.36),p=0.031),syncope[HR:3.09(1.45-6.58),p=0.003]and LGE extent(per 10%)[HR:1.65(1.26--2.17),p<0.001]were independent predictors for unfavorable prognosis in HCM.Conclusions This study showed that the HCM phenotype is related with certain CMR parameters:left ventricular maximal wall thickness,the shape of ventricular septum and LGE extent.There were no significant differences in the CMR parameters and prognosis in patients with MYBPC3 and MYH7 mutations.LGE extent was an independent predictor of cardiovascular events in HCM patients,while the positive genotype was associated with unfavorable prognosis in HCM but was not an independent predictor.