CMR-derived ECVs vary with myocardial region and associate with the regional wall thickness

This study was designed to identify whether the position and size of the region of interest (ROI) influence extracellular volume fraction (ECV) measurements. Patients with localized (n = 203) or infiltrative (n = 215) cardiomyopathies and 36 normal controls were enrolled in this study. ECV measurements at 4 different regions, including the anterior, septal, posterior and lateral wall regions, were measured, and all groups were compared. Regional ECV was correlated with the corresponding regional wall thickness. The diagnostic power to differentiate the myocardial abnormalities was evaluated for each myocardial region. ECVs measured using five different ROI sizes within each myocardial region were compared. Our results showed that ECVs varied among the myocardial regions, and this variation was significantly associated with regional wall thicknesses. For the detection of myocardial abnormalities, regional ECV revealed similar results as ECV derived from the whole region except for the anterior region. No significant difference was found in the ECVs measured using the five different ROI sizes. In conclusion, CMR-derived ECVs vary with myocardial region, and this variation is significantly associated with the regional wall thickness. In contrast, the measured size of the ROI has less of an effect on the ECV.


Results
Patient characteristics. The demographics of the study population are summarized in Table 1. Compared with the control group, the patient groups were older and included more males; however, body surface areas were similar. There were significant differences in LV function and mass at end-diastole between the patient and control groups. In the patient group, ten different types of cardiomyopathies were included for further analysis. The homogeneous disease (H.D.) group (n = 215) included cardiac amyloidosis (n = 15), arrhythmogenic right ventricular cardiomyopathy (n = 47), Brugada syndrome (n = 64), dilated cardiomyopathy (n = 37), Fabry disease (n = 24) and heart failure with preserved ejection fraction (n = 28). The regional disease (R.D.) group (n = 203) included hypertensive cardiac disease (n = 81), hypertrophic cardiomyopathy (n = 87), ischemic cardiomyopathy (n = 21) and myocarditis (n = 14).
Interregional variation of ECV and its relationship with regional wall thickness. The inter-ECV within four different regions was significantly higher in the patient group than in the control group (compared with R.D. group: 2.13 ± 1.16% vs. 1.29 ± 0.52%, p < 0.001; compared with H.D. group: 1.82 ± 0.93% vs. 1.29 ± 0.52%, p < 0.001). In addition, the inter-ECV measured in the R.D. group was significantly higher than that measured in the H.D. group (2.13 ± 1.16% vs. 1.82 ± 0.93%, p = 0.016) (Fig. 1b). A relatively broad range of inter-ECV was found in the R.D. group compared with the H.D. (range 0.26-6.68% vs. 0.09-5.54%) and control groups (range 0.26-6.68% vs. range 0.38-2.77%). In addition, Fig. 5 showed that the ECV reg was significantly correlated with the corresponding regional wall thickness in the patient cohort, except for patients with Fabry Figure 1. Group comparisons of the whole regional extracellular volume fraction (ECV) (a) and the interregional variation of ECV among the groups (b). ECV whole the whole regional ECV; inter-ECV the interregional variation of ECV; R.D. regional disease; H.D. homogeneous disease; Ctr. control. www.nature.com/scientificreports/ disease and myocarditis (Table 4). This significant correlation was also found in the control groups (r = 0.183; p = 0.028) (Fig. 6). Table 5 compares the ECV results derived from

ECV measurements from ROIs of different sizes.
ROIs of different sizes in the patient and control groups. No significant difference was found in the ECV measured from different sized ROIs for each myocardial region in either the patient or control group.
Reproducibility. For the intra-observer variability, the ECV of repeated measurements range from 19.3% to 49.4% with medium 24.8%. The bias was 0% with 1.048% standard deviation of the absolute difference (Fig. 7a).
For the inter-observer variability, the ECV of repeated measurements range from 19.3% to 49.4% with medium 24.5%. The bias was -0.131% with 1.184% standard deviation of the absolute difference (Fig. 7b).

Discussion
In this study, we demonstrated that CMR-derived ECVs varied with the myocardial region and that this variation was significantly increased in patients with ventricular hypertrophy. Our results also showed that the diagnostic power of regional ECVs was similar to that of ECVs derived from the whole region except for the anterior region. These findings suggest that ECV measurement using the ROI-based method is feasible to detect myocardial abnormalities compared with the measurement from the whole region. However, regional ECV may lead to clinically significant errors if ECV was assessed and compared in different myocardial regions. In contrast, no significant difference was found in the ECVs from the 5 different ROI sizes. This finding suggests that the size of the ROI has less of an effect on the ECV measurement based on the ROI-based method. CMR-derived ECV is a useful biomarker to quantify diffuse interstitial matrix 2,5,8,12 and predict adverse cardiovascular outcomes [13][14][15] . ECV is most often measured using an ROI-based method, which assumes that the myocardial abnormality is homogeneously distributed (uniform ECV) within the noninfarcted myocardial regions. Kawel et al. 16 used CMR to evaluate acquisition factors that may result in the variation of measured ECV, including magnetic field strength, cardiac phase and myocardial region in normal controls. They found that ECV did not vary significantly with field strength but did vary with cardiac phase and myocardial region. They concluded that ECV was significantly higher for the septum compared to the nonseptal myocardium in normal controls. Recently, Vita et al. demonstrated that ECV varies according to the myocardial region in patients with non-ischemic dilated cardiomyopathy 17 . In this study, we further demonstrated that this interregional variation of ECV was more significant in the patient group than the control group and was positively associated with the corresponding regional wall thickness. These findings suggest that the degree of regional variation in ECV might be associated with the severity of ventricular hypertrophy.
Several segmentation approaches are used to quantify the ECV in clinical settings, including the entire LV myocardium, the septal myocardium and ROI in the noninfarcted myocardial region 8 . Since CMR-derived ECV values may vary with myocardial region, it is essential to evaluate whether the position of the ROI affects the ability to differentiate myocardial abnormalities. In this study, we used ROC analysis to assess the diagnostic www.nature.com/scientificreports/ power of ECV measurements from each region and from the whole region of myocardium. Our results showed that the diagnostic power using regional ECV measured from the septal, posterior and lateral regions was similar to that measured from the whole region. This finding suggested that the ECV measurements from these regions can feasibly differentiate the myocardial abnormalities compared with the measurements from the whole myocardium. Our findings also indicated that ECV measurements from the anterior region had lower distinguishing power than the other regions. In this cohort, the SD was 0.82% for the ECV abnormality cut-off values in each myocardial region. For the intra-and interobserver variability tests, the bias SDs were 1.048% and 1.184%, respectively. Therefore, regional ECV variations did not lead to reclassification of the myocardial region being normal or abnormal, and no need to define the various cut-off values depending on myocardial regions. CMR-derived ECV can be evaluated using the ROI-based method either from native and postcontrast T1 maps or directly from ECV mapping 6 . In this study, we demonstrated that CMR-derived ECV measurements had significant interregional variation in both the control and patient groups. These variations were positively associated with the regional wall thickness. Therefore, caution must be exercised in estimating the ECV in patients with significant ventricular hypertrophy. The position of the ROI should be obtained in the same region or should cover the entire LV myocardium. In addition, our results showed that there was no significant difference in the ECVs from these five different sized ROIs. Nonetheless, we noted that the SD of the ECV within the ROI was lower in larger ROIs than in smaller ROIs. This finding may imply that one should select as large an ROI as possible without including ventricular blood and papillary muscle.
There are several limitations to this study. First, our study was done with a single T1 pulse sequence (MOLLI). Different sequences of T1 mapping have been reported to yield different absolute ECV values 18 . The inter-ECV results were based on the relative difference in ECV among the four regions. Whether this variation is identical in different pulse sequences is unknown and needs further investigation. Second, our results indicated that the control group had two regional variations in ECV compared with four in the patient group (Fig. 2). The control group was smaller than the patient group, which may have limited our ability to detect subtle differences between regions. Whether the results of this study were merely a matter of sample size remains unknown. Nonetheless, there appears to be the same tendency of septal ECV being larger than the other regional ECVs in all groups. Third, although all subjects were carefully controlled in data acquisition and analysis, our results do not have www.nature.com/scientificreports/ histological evidence to support the suggestion that this interregional variation of ECV reflects a pathophysiological difference rather than a technical difference. Forth, regional variation of ECV was only performed in one representative mid-ventricular slice, future studies should consider apical and basal slices for more comprehensive assessments.
In conclusion, our study demonstrated that CMR-derived ECVs vary with myocardial region in both the patient and control groups. This interregional variation is associated with the severity of ventricular hypertrophy. Regional ECV is feasible for differentiating myocardial abnormalities compared with ECV derived from the whole region except for in the anterior region. When conducting ECV measurements using the ROI-based method, the position of the ROI is essential for comparing ROIs within the same region, but the measured size of the ROI is less critical to consistencies in ECV measurement.

Ethics statement. The research was approved by the institutional review board of the National Taiwan
University Hospital Ethics Committee. The study was conducted in accordance with the approved guidelines. All study participants provided written informed consent.

Study population.
Between January 2017 and October 2019, 647 subjects undergoing clinical or research CMR were retrospectively recruited for T1 mapping with the calculation of ECV. Exclusion criteria included subjects with suboptimal image quality due to arrhythmia (n = 34), examination without intravenous contrast administration (n = 22), and LGE images with hyperenhancement (n = 137). The final cohort consisted of 418 patients and 36 normal controls. Patients were divided into two groups, the regional disease group (R.D. group)  Table 3. The extracellular volume fraction (ECV) cut-off values determined from the sensitivity and specificity of each myocardial region. www.nature.com/scientificreports/ Figure 5. Linear regression between the regional ECV (ECV reg ) and the corresponding regional wall thickness (WT reg ) in the patient cohort. ARVC arrhythmogenic right ventricular cardiomyopathy; BrS Brugada syndrome; DCM, dilated cardiomyopathy; Fabry Fabry disease; HCM hypertrophic cardiomyopathy; HTC hypertensive cardiac disease; HFpEF heart failure with preserved ejection fraction; ICM.

Cut-off value (%) Sensitivity (%) Specificity (%)
Scientific Reports | (2020) 10:20965 | https://doi.org/10.1038/s41598-020-78043-5 www.nature.com/scientificreports/  Figure 6. Linear regression between the regional ECV (ECV reg ) and the corresponding regional wall thickness (WT reg ) in the control group. Imaging analysis. Commercial postprocessing software (cvi42, Circle Cardiovascular imaging, Calgary, AB, Canada) was used to analyze the LV function, regional wall thickness and ECV offline. Maximum wall thickness was measured from end-diastolic short-axis cine image. The wall thickness of the anterior, septal, posterior and lateral segments was measured at the same level of the T1 mapping. The ECV was calculated from native and postcontrast T1 maps using a region-based method and was then calibrated with the last available hematocrit data. Four different regions, including the anterior, septal, posterior and lateral wall regions of the LV, and the area in the central area of the LV cavity were drawn on the T1 map at the mid-ventricular slice. The position of the ROI was carefully drawn to avoid the inclusion of trabeculation, blood or pericardium in the thin-walled parts of the myocardium. The regional ECVs (ECV reg ) measured from four different regions and the whole regional ECV (ECV whole ) calculated from the four ECV reg were compared in each group and among the groups. In addition, the interregional variation of ECV (inter-ECV) was assessed by measuring the standard deviation (SD) of ECV reg among these four regions. We hypothesized that the variations in ECV might be associated with the ventricular wall thickness. The ECV reg for all subjects were correlated with the corresponding regional wall thickness. Five different sized ROIs drawn within the myocardial region were selected to quantify the ECV in 90 subjects who were randomly selected from each group. To standardize the analyzed procedures, five 3 × 3 ROI pixels were placed equally across the myocardial regions. Five different ECV sizes were measured from one to five ROI 3 × 3 pixels. To evaluate the reproducibility of the ECV measurements, intra-and interobserver variabilities were studied in 120 myocardial regions of 30 subjects randomly selected from each group.

Statistical analysis.
All continuous data were first tested for fitting the normal distribution using the Shapiro-Wilk test. Continuous variables were expressed as the means and SD, and categorical variables were expressed as the percentages. Since the Shapiro-Wilk test showed that most ECV values were not normally distributed, all statistical analyses were performed using nonparametric methods. Group comparisons between the measured ECV in each region were analyzed with the Mann-Whitney U test. Regional comparisons among the measured ECVs for each group were performed by using the Kruskal-Wallis test, and the Mann-Whitney U test was used for post-hoc analysis. The statistical tests were two-tailed, and statistical significance was defined as P < 0.05. Spearman's rank correlation was used to obtain correlation coefficients between continuous variables of interest. Agreements between the measurements were assessed via the intraclass correlation (ICC) coefficient uing a two-way random-effects model. The receiver operating characteristic (ROC) analysis was performed Figure 7. Intra-observer (a) and inter-observer (b) variabilities of regional ECV measurements from 120 myocardial regions in 30 subjects randomly selected from each group.