The Genetic Characteristics Based On Next Generation Sequencing And Association With Cardiac MRI Depth Phenotyping In Hypertrophic Cardiomyopathy

Background and Objective:Hypertrophic cardiomyopathy(HCM)is a common hereditary cardiomyopathy.The study of phenotypic and genetic characteristics can help more accurately individualized classification and risk stratification to guide clinical decision-making.Next-generation gene sequencing,as the current mainstream gene sequencing tool,provides a more comprehensive technology for us to understand the genetic characteristics of HCM.In addition,cardiac magnetic resonance imaging(MRI),as a new imaging tool for the evaluation of HCM,also plays an important role in the diagnostic and prognostic evaluation.However,little study exists on the genetic characteristics in Chinese population with HCM patients and the genotype-phenotype association as assessed by cardiac MRI.This study uses the technical advantages of cardiac MRI,next generation sequencing,and machine learning algorithms in a prospective HCM cohort to explore the genetic characteristics of the Chinese HCM population,as well as the relationship between genotype,and clinical diagnostic phenotype or cardiac MRI-phenotype.When the existing diagnostic phenotype and other conventional MRI-phenotype data cannot obtain evidence of genetic etiology,we have innovatively explored whether the use of radiomics method or the addition of new imaging marker can improve the genetic diagnosis of HCM and the ability to predict adverse events,or provide corresponding evidence for the pathogenic mechanism of HCM,and further provide scientific basis for elucidating the genetic epidemiological characteristics,cardiac MRI-phenotypes,and interrelationships of HCM in the Chinese population.Materials and Methods:Part 1:We prospectively enrolled 383 unrelated HCM probands who were treated in West China hospital of Sichuan University from 2015 to 2019 and received genetic testing and 3.0 T cardiac MRI examination,including 118 familial HCM and 265 sporadic HCM.The clinical phenotypic diagnosis of HCM is based on the diagnostic criteria of the American Heart Association and American College of Cardiology guidelines.Patients with familial HCM underwent targeted next generation sequencing including 117 candidate pathogenic genes associated with cardiomyopathies,while sporadic HCM underwent whole-exome sequencing.The genotypic characteristics were analyzed,and the relationship between the genotype and clinical diagnostic phenotype of the included cohort was explored.Part 2:To further explore the relationship between genotype and cardiac MRIphenotype,we prospectively enrolled 144 HCM patients with pathogenic or likely pathogenic gene mutations who underwent genetic testing and 3.0 T cardiac MRI from 2016 to 2017.MRI cine,late gadolinium enhancement(LGE)and T1 mapping images were acquired.The genotype and cardiac MRI phenotypic characteristics including cardiac function,the distribution and percentage of LGE,native T1 and extracellular volume(ECV)were analyzed for exploring the association of genotype and cardiac MRI-phenotype in hypertrophic patients with different sarcomere-related gene mutations and some rare gene mutations.In order to explore whether radiomics features combined with machine learning algorithms can improve the ability of HCM genetic diagnosis,a total of 157 quantitative features were extracted from T1 Mapping images of 68 patients with MYH7 gene-related and 34 MYBPC3 gene-related HCM patients,and the patients were randomly assigned to the feature selection and test validation group at a ratio of 4:1.Then the two most common genotypes in HCM(MYH7 and MYBPC3)are identified through the extracted radiomics features and the established classifier by support vector machine(SVM)combined with principal component analysis(PCA).Furthermore,receiver operating characteristic(ROC)curve analysis is used to evaluate the effectiveness of the classification model.Part 3:We prospectively enrolled 378 unrelated HCM patients who underwent 3.0 T cardiac MRI examination from August 2011 to October 2017.All MRI cine and LGE images were acquired.We also recruited 100 healthy participants of ageand sex-matched to form a normal control group.Fractal analysis(FD)of MRI cine images was performed to quantify the characteristics of trabeculae complexity in all participants.We set the normal reference value of the left ventricular FD according to the upper limit of the normal reference value of the healthy group(mean+2 standard deviation),and then use this reference value to divide HCM patients into increased and normal FD HCM subgroups.All HCMs were followed up for clinical adverse events.The primary endpoint was defined as all-cause mortality and aborted sudden cardiac death,and the secondary endpoint was the composite event of the primary endpoint and heart failure readmission.The survival curve was drawn based on the Kaplan-Meier method,and the log-rank test was used to compare the risk probability of adverse events in the increased and normal FD HCM subgroups.The Cox proportional hazards model was used to test whether the LV FD measurement is an independent predictor of the primary and secondary endpoints and to evaluate its prognostic performance.Finally,internal validation of the model is carried out by using 100 Bootstrapping method.In addition,188 unrelated HCM patients who underwent genetic testing were analyzed for the relationship between genotype and trabecular complexity.Results:Part1:Among the 383 unrelated Chinese HCM probands,we reported 56 new pathogenic or likely pathogenic gene site mutations,including 3 new rare non-sarcomeric gene mutations and 53 new sarcomeric gene mutations.According to the analysis of genotypic characteristics,it was found that the frequency of rare genes was 3.4%.After excluding rare diseases,the gene-positive diagnosis rate was 62.4%.Among them,MYH7 and MYBPC3 are the most common pathogenic genes,accounting for 61.9%of patients with positive genotype.Additionally,in the included HCM population,there was no significant difference in the gene-positive diagnosis rate between familial HCM and sporadic HCM(66.1%vs 64%,P=0.7).The proportion of sporadic HCM carrying polygenic mutations was higher than that of familial HCM(11.2%vs.1.8%,P=0.003).Part 2:The 7 cases of glycogen storage disease subgroup had higher LV mass(99.6 g/m2,interquartile range(IQR):81.1-290.2 g/m2 vs.71.5 g/m2,IQR:59.7-94.7 g/m2;P=0.03),diffuse LGE pattern,higher LGE%and native T1(1345.6 ms,IQR:1342.1-1367.4 ms)vs.1280.7 ms,IQR:1236.5-1312.9ms;P=0.002)compared to the 137 cases of HCM subgroup with sarcomere protein gene mutations.However,the baseline(age,gender,body mass index,and blood pressure)and cardiac MRI characteristics(LGE%,native T1 and ECV)between 19 cases of thin-filament and 102 cases of thick-filament HCM subgroup or the 68 cases of MYH7 and 34 cases of MYBPC3 HCM subgroups was similar(all P>0.05).Furthermore,it was found that the extracted T1 mapping radiomics features and the established SVM combined with PCA classifiers can be used to distinguish the two most common genotypes of HCM(MYH7 and MYBPC3).The accuracy and area under the curve(AUC)are respectively 92.0%and 0.968(95%confidence interval(CI):0.968-0.971).For the test validation data set,the accuracy and AUC are 85.5%and 0.886(95%CI:0.881-0.901),respectively.Part 3:During the median follow-up of 33±18 months,HCM patients with left ventricular maximum apical FD(≥1.325)had a higher risk of primary endpoint(P=0.01)and secondary endpoint(P=0.04).In addition,Cox regression analysis showed that after individually adjusting the predictor variables of European Society of Cardiology(ESC)risk model,and LGE%,the left ventricular maximum apical FD is still the only independent predictor in predicting the primary endpoint(hazard ratio range,1.001-1.008;all P<0.05)and secondary endpoint events(hazard ratio range,1.006-1.007;all P<0.05)in patients with HCM.Internal verification showed that the left ventricular maximum apical FD still maintained an excellent ability to predict the primary endpoint event in the internal validation,and the C index was 0.70±0.03.In addition,compared with healthy volunteers,the left ventricular FD measurements of both gene-positive and gene-negative HCM patients were significantly higher(LV maximal apical FD:1.35±0.06(gene-positive HCM group)and 1.35±0.06(gene-negative HCM group);compared with 1.27±0.03(healthy control group);P<0.001).However,there was no significant difference in trabecular complexity between gene-positive and gene-negative HCM,and no gene-specific differences(MYBPC3 vs.MYH7 mutation,maximum apical FD:1.35±0.07 vs.1.36±0.05;P=0.337;thick filament vs.thin filament mutation,maximum apical FD:1.35±0.06 vs.1.29±0.08;P=0.052).Conclusion:First,we explored the genetic heterogeneity of HCM based on the clinical diagnostic phenotype of HCM and found 56 new pathogenic or likely pathogenic gene mutations,which expanded the gene spectrum of HCM.In addition,the frequency of rare disease genes in the HCM cohort of the Chinese population is 3.4%,and the composition ratio of genotypes is similar to that of foreign HCM populations.The gene positive rate of sporadic HCM is similar to that of familial HCM,but the composition ratio of genotypes is different.Second,there are differences in cardiac MRI characteristics such as LGE and native T1 values between patients with glycogen storage disease and HCM with mutations in the sarcomere protein gene.Third,there were no significant conventional MRI phenotypic differences between HCM with mutations in thick and thin filaments,or between MYH7 and MYBPC3.However,the T1 mapping image features extracted by radiomics analysis combined with machine learning algorithms can distinguish the most common genotypes of HCM(MYH7 and MYBPC3).Fourth,the left ventricular maximum apical fractal dimension(FD)in HCM has independent predictive value for all-cause death and composite adverse events.In addition,our research has also found that there is no significant difference of LV FD between gene-positive and gene-negative HCM patients and no gene-specific differences,indicating that the formation of trabecular muscles may be caused by the interaction of genetic defects and environmental factors in patients with HCM.