Evaluation Of Doxorubicin-induced Cardiotoxicity And Myocardial Fibrosis Using Cardiac Magnetic Resonance Imaging

Objective:Doxorubicin（DOX）is a commonly used anthracycline chemotherapy drug.Cardiotoxicity is the most serious adverse reaction during the use of DOX,which can cause myocardial fibrosis,and lead to dilated changes in the myocardium and eventually develop into heart failure.Therefore,effective evaluation of doxorubicin-related cardiotoxicity and myocardial fibrosis is very important for monitoring the side effects of doxorubicin.In the first chapter of this study,dynamic cardiac magnetic resonance scanning was performed on the pathogenesis of DOX-induced cardiomyopathy（DIC）in rats,and various post-processing techniques were used to extract magnetic resonance imaging parameters to longitudinally study the pathogenesis of doxorubicin-induced rat cardiomyopathy and to explore the early imaging indicators of doxorubicin-related cardiotoxicity.At the same time,we also explored to construct a predictive model for DIC in rats.In addition,the second chapter of this study aims to construct an evaluation model of myocardial fibrosis with reference to the pathological standards using cardiac magnetic resonance imaging technology.Materials and Methods:The first chapter of our study included thirty-nine male Sprague Dawley（SD）rats.They were divided into an experimental group（n=24）and a control group（n=15）.The experimental group were injected with doxorubicin through vena caudalis weekly to induce cardiomyopathy model（2.5mg/kg,1 time/week,6 times in total）.CMRI was performed with a 7.0T small animal MRI scanner once a week for 9 times in total.After the last MRI scan,the rats were sacrificed and the heart was taken out for pathological analysis.Post-processing of cardiac magnetic resonance images to obtain routine left ventricular function parameters,whole left ventricular wall thickness parameters,whole myocardial strain parameters,normalized left ventricle myocardial T2 value.The wall thickness parameters and myocardial strain parameters of 16segments（not included in the 17th segment）were obtained,and a total of 104parameters were obtained.The differences of these parameters between the two groups were analyzed,and the early imaging indicators of doxorubicin-related cardiotoxicity were screened.The correlation analysis between myocardial strain,wall thickness,normalized myocardial T2 value and LVEF were performed.Then,stepwise multivariate Logistic regression analysis was performed on the magnetic resonance parameters which had good correlation with the LVEF to construct the predictive diagnosis model of doxorubicin-induced cardiomyopathy in rats.Fifty male SD rats were included in the second chapter of this study.Thy were divided into the experimental group（n=30）and the control group（n=20）.The experimental group and the control group were divided into 5 subgroups.Each subgroup of the experimental group includes 6 rats,and each subgroup of the control group includes 4 rats.The rats in the experimental group were injected with doxorubicin through vena caudalis weekly to induce myocardial fibrosis model（2.5mg/kg,1 time/week,6 times in total）.In the 5 subgroups of the experimental group,the number of injections were 2,4,6,6,and 6,respectively.The time between the MRI scan and the first injection was 2 weeks,4 weeks,6 weeks,8 weeks and 10weeks respectively.The 5 subgroups of the control group were injected with saline and underwent MRI scan at the time corresponding to the experimental group.After each MRI scan,the rats were sacrificed and the heart was taken out for pathological analysis,and the degree of myocardial fibrosis was quantified.The 103 MR parameters were obtained following the method in Chapter 1,including routine left ventricular function parameters,whole and segmental wall thickness parameters,whole and segmental myocardial strain parameters.Correlation and stepwise multivariate regression analysis of imaging parameters and pathological results were carried out to construct a assessment model of myocardial fibrosis.The prediction equation of DIC occurrence probability can be obtained by transforming the equation of Logistic regression modelResults:In the first chapter of this study,after the modeling,the left ventricular ejection fraction（LVEF）and left ventricular myocardial mass（LVM）in the experimental group were lower than the control group（54.745±3.960%vs 66.869±3.474%,P<0.001;0.358±0.054g vs 0.589±0.044g,P<0.001）.The left ventricular end-diastolic volume（LVEDV）（0.541±0.073m L vs 0.448±0.064m L,P<0.001）and left ventricular end-systolic volume（LVESV）（0.207±0.029 m L vs 0.165±0.039m L,P<0.001）of rats in the experimental group were larger than those of the control group.Among the above parameters,LVEDV,LVESV,and LVEF all decreased at the 9th CMRI,while LVM began to decrease at the 5th CMRI.In the overall wall thickness parameters,except for the middle layer of the end systolic wall thickness,other wall thickness parameters were changed compared with the control group after modeling.Moreover,the time for the thinning was mostly during the 9th CMR scan,and the end-diastolic wall thickness at the basal layer and the end-systolic wall thickness at the apical layer showed thinning during the 8th CMR scan.The time when the wall thickness of some segments became thinner was also at the 8th CMR scan,while the time when the thickness of the segment 7 became thinner was at the 7th CMR scan.The overall and segmental myocardial strain parameters of the experimental group were lower than those of the control group,and the overall myocardial stress parameters began to decrease at the 7th MRI scan.The LVGPRS,LVGPCS,and the left ventricular middle peak radial decreased earlier,and they began to decrease at the6th MRI scan.In the segmental myocardial stress parameters,the segmental peak radial strain began to decrease at the 6th and 7th MRI scans,and the peak radial strain of segment 2 began to decrease at the 5th MRI scan;The peak value of segmental circumferential strain mostly began to decrease at the 7th MRI scan or later;The peak longitudinal strain of the segment began to decrease after the 8th MRI scan.The standardized myocardial T2 values of the experimental group were higher than those of the control group（0.820±0.069 vs 0.745±0.113,P=0.014,respectively）at the 6^th（5 weeks after administratio）and 7^th MRI scans（6 weeks after administration）（0.820±0.069 vs 0.745±0.113,P=0.014;0.734±0.086 vs 0.919±0.114,P<0.001）.There was no statistical difference in the standardized myocardial T2 values of rats in the experimental group compared with the control group at the other time points of MRI scanning.The absolute values of correlation coefficients between most wall thickness parameters and LVEF were less than 0.3,so the correlation between the wall thickness and LVEF was weak.Most of the myocardial strain parameters of the left ventricle were correlated with LVEF,and the absolute values of the correlation coefficients were greater than 0.5.The correlation coefficients between myocardial strain parameters and LVEF were generally higher than correlation coefficients between wall thickness and LVEF.We selected 40 myocardial strain parameters（12 parameters of whole myocardial strain,10 parameters of segmental radial myocardial strain,9parameters of segmental circumferential myocardial strain,9 parameters of segmental longitudinal myocardial strain）based on the correlation analysis to build the predictive diagnosis model of doxorubicin-induced cardiomyopathy（DIC）.Stepwise multivariate Logistic regression analysis was used to construct the predictive diagnostic model of DIC.The whole myocardial strain parameters for the DIC diagnostic model contains three strain parameters,the segmental radial myocardial strain parameters for the DIC diagnostic model contains four strain parameters,the segmental circumferential myocardial strain parameters for the DIC diagnostic model contains three strain parameters,and the segmental longitudinal myocardial strain parameters for the DIC diagnostic model contains three strain parameters.Among these models,the predictive model of DIC constructed by segmental radial myocardial strain parameters had the highest value（pseudo R²=0.806,correctly classified=0.972,sensitivity=0.875,specificity=0.984,AUC=0.992）.In the second chapter of this research,the degree of myocardial fibrosis in subgroup 3,4 and 5 of the experimental group were higher than those of the corresponding control group（8.784±1.829%vs 4.566±1.550%,P=0.005;15.861±1.805%vs 4.415±1.477%,P<0.001;22.368±2.146%vs 4.052±0.917%,P<0.001）。Among the routine cardiac function parameters,left ventricular ejection fraction（LVEF）,left ventricular myocardial mass（LVM）,left ventricular end-diastolic volume（LVEDV）,and left ventricular end-systolic volume（LVESV）were all correlated with the degree of myocardial fibrosis.Among the myocardial fibrosis evaluation models constructed by these parameters,the model constructed by LVEF,LVEDV and LVM had the best explanatory power and fitting degree（R²=0.847,AIC=242.794）.The parameters of the whole and segmental wall thickness of the left ventricle were mostly correlated with the degree of myocardial fibrosis.Among the myocardial fibrosis evaluation models constructed by these wall thickness parameters,the model constructed by LVESWT＿AHA6,LVESWT＿AHA14 and LVESWT＿AHA1 had the best explanatory power and fitting degree（R2=0.745,AIC=268.365）.Both the whole and segmental myocardial strain parameters were correlated with the degree of myocardial fibrosis.Among the myocardial fibrosis evaluation models constructed by these myocardial strain parameters,the evaluation model constructed by the whole myocardial strain parameters has better explanatory power and fitting degree than the models constructed by segmental radial,circumferential and longitudinal strain parameters（R² were 0.865,0.852,0.815,0.848,and AIC were 240.027,246.558,253.349,249.643,respectively）.In addition,some routine cardiac function parameters were combined with myocardial strain parameters to obtain the best model for evaluating the degree of myocardial fibrosis.The magnetic resonance parameters included in this model were LVEF,LVEDV,LVBPCS,LVPCS＿AHA4 and LVPCS＿AHA5.The R² of this model was 0.953,and the AIC was 187.925.Conclusion:Cardiac magnetic resonance imaging can effectively evaluate the structure and function of doxorubicin-induced cardiomyopathy（DIC）in rats.Whether the wall thickness and strain in each segment of the DIC model change and when they change are inconsistent,suggesting the heterogeneity of myocardial damage in rats with DIC.Because doxorubicin-related cardiotoxicity is the mechanism that induces DIC,the imaging indicators that change in the early stage of modeling are also early imaging indicators of doxorubicin-related cardiotoxicity.These indicators include LVM,standardized myocardial T2 value,some myocardial strain parameters（especially segmental radial strain parameters）.Since this study is a prospective longitudinal study of the pathogenesis of DIC in rats,some myocardial strain parameters have changed before the occurrence of DIC and had high correlation with left ventricular ejection fraction.Therefore,these indicators may have a certain predictive effect on the occurrence of cardiomyopathy.In this study,the predictive model of DIC constructed with these parameters can well predict the occurrence of DIC（In particular,the model based on segmental radial strain parameters has the highest value of diagnosis）,which is conducive to clinical monitoring of the side effects of doxorubicin.Myocardial fibrosis is the important pathological change of DIC.The cardiac function parameters,wall thickness parameters and myocardial strain parameters obtained from cardiac magnetic resonance can non-invasively assess the degree of myocardial fibrosis.The assessment model of myocardial fibrosis constructed by the myocardial strain parameters has good explanatory power,espically the model constructed by whole myocardial strain parameters.