Evaluation of extracellular volume by computed tomography is useful for prediction of prognosis in dilated cardiomyopathy

Cardiac computed tomography (CT) is useful for the screening of coronary artery stenosis, and extracellular volume fraction (ECV) analysis by CT using new dedicated software is now available. Here, we evaluated the utility of ECV analysis using cardiac CT to predict patient prognosis in cases with dilated cardiomyopathy (DCM). We analyzed 70 cases with DCM and cardiac computed tomography (CT) with available late-phase images. We evaluated the ECV of the left ventricular myocardium (LVM) using commercially available software (Ziostation 2, Ziosoft Inc, Japan). ECV on LVM was 33.96 ± 5.04%. Major adverse cardiac events (MACE) occurred in 21 cases (30%). ECV of the LVM on CT, endo-systolic volume, and rate of significant valvular disease were significantly higher in cases with MACE than in those without (37.16 ± 5.91% vs. 32.59 ± 3.95%, 194 ± 109 vs. 138 ± 78 ml and 57% vs. 20%, all P values < 0.05). LVEF was significantly lower in cases with MACE than in those without (23 ± 8 vs. 31 ± 11%, P = 0.0024). The best cut-off value of ECV on LVM for prediction of MACE was 32.26% based on receiver operating characteristics analysis. Cases with ECV ≥ 32.26% had significantly higher MACE based on Kaplan–Meier analysis (P = 0.0032). Only ECV on LVM was an independent predictor of MACE based on a multivariate Cox proportional hazards model (P = 0.0354). Evaluation of ECV on LVM by CT is useful for predicting MACE in patients with DCM.


Background
Dilated cardiomyopathy (DCM) is commonly encountered in daily clinical practice. It is the third-most common reason for heart failure and has an estimated prevalence of one in 2500 in the average population [1]. Diagnosis is dependent on screening for coronary artery disease, other cardiomyopathies and other conditions causing abnormal loading, including valvular heart disease and hypertension [1][2][3].
The gold standard for evaluation of myocardial damage is late enhancement analysis using cardiac magnetic resonance imaging (MRI). The pattern of late gadolinium enhancement (LGE) is useful in the differential diagnosis of several types of cardiomyopathies [4]. Linear mid-layer LGE is the typical pattern in patients with DCM, and the presence of the LGE has been regarded as a sign of worse prognosis in these patients [5,6].
Of interest, extracellular volume fraction (ECV) analysis using T1 mapping images on MRI is also useful for quantitative analysis of myocardial damage. This analysis requires gadolinium contrast [7]. ECV provides significant prognostic information in DCM, and patients with adverse events have significantly higher ECV than those without [8].
Cardiac MRI is contraindicated in patients with implanted mechanical devices, claustrophobia or severe renal dysfunction, and obtaining clear LGE images in cases with arrhythmia is often difficult [9]. Cardiac computed tomography (CT) is useful for screening coronary artery stenosis in patients suspected of having myocardial disease [10], and obtaining clear cardiac images is easy even in cases with arrhythmia [11]. Additional late phase scan is also helpful for detecting myocardial damage as late enhancement [12]. New software has recently allowed ECV analysis on CT, and ECV values on CT are similar to those on MRI [13]. To our knowledge, however, the usefulness of ECV analysis on CT for predicting patient prognosis in DCM has not been reported.
The purpose of this study was to evaluate the utility of ECV analysis on CT in predicting major adverse cardiac events (MACE) in patients with DCM.

Methods
Ninety-three patients diagnosed with DCM underwent cardiac computed tomography (CT), including late-phase acquisition, in our institution from January 2009 to March 2022. However, ECV analysis was impossible because of significant metallic artifacts of pacemaker leads in four patients, significant gaps in cardiac phases between the non-contrast and late phase cardiac images in two patients, and the different tube voltages between the non-contrast and late phase cardiac images in 17 patients. The study was conducted under a retrospective design in the remaining 70 consecutive patients with DCM. All patients had a lower left ventricular (LV) ejection fraction (LVEF) less than 45% on transthoracic echocardiography (TTE) [2]. Exclusion criteria were as follows: significant coronary artery stenosis on cardiac CT, moderate or severe primary mitral or aortic valve disease, ischemic late enhancement pattern of left ventricular myocardium (LVM) on cardiac CT, clinical history of myocarditis, and end-stage hypertrophic cardiomyopathy [14,15]. Cardiac CT was performed to screen for coronary artery disease (CAD), but no significant coronary artery stenosis was indicated in any patient. A significantly stenotic coronary artery was defined as a coronary artery with ≥ 70% stenosis (a left main coronary artery with > 50% stenosis) on cardiac CT [16]. Major adverse cardiac events (MACE) were defined as a composite of cardiovascular death, fatal arrhythmic events, and hospitalization due to heart failure. Follow-up information was obtained from patient medical records at the institution. Patient background, including risk factors for coronary artery disease and medical treatment, were obtained from medical records. The institutional review board approved this retrospective analysis (Reference No. 3822), and the study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

Protocol for computed tomography
CT was performed using a 320-slice CT (Aquilion One or Aquilion One/ViSION Edition, Canon Medical Systems, Otawara, Japan) or 256-slice CT (Revolution Apex, GE Healthcare, Milwaukee, WI), with patients lying supine on the scanner table. After a scout scan, a non-contrast ECGgated cardiac scan was performed using a prospective ECGgated technique targeting the diastolic phase. Slice thickness and tube voltage were 0.5 mm and 80-120 kV for 320-slice CT and 0.625 mm and 70 kV for 256-slice CT, respectively.
For retrospective ECG gating, performed where possible using the dose modulation technique to decrease radiation dose during the systolic phases, conventional enhanced CT was performed with a slice thickness and tube voltage of 0.5 mm and 120 or 135 kV for 320-slice CT and 0.625 mm and 120 kV for 256-slice CT, respectively [17]. Tube current at scanning was determined based on the auto exposure control system. All patients with a heart rate ≥ 65 beats per minute received 10 mg of propranolol or 12.5 mg landiolol prior to scanning, except for those in whom β-blockers were contraindicated. Just prior to the scanning procedure, subjects were administered two doses of isosorbide dinitrate sublingually.
For contrast material injection, we employed a routine triphasic protocol because we previously identified the protocol for higher diagnostic accuracy of detection of LE on CT [12,18,19]. Right or left antecubital intravenous access using a 20-or 22-gauge needle was attained, and the system was connected to a dual-syringe injector with a dual-flow option (Dual Shot, Nemoto, Tokyo, Japan). During the first phase, we injected 45-100 ml of undiluted iodinate contrast agent (350-370 mg/ml) at 3-5 ml/s, followed by 0-70 ml of a 50/50% saline-to-contrast material mixture at 2-4 ml/s and 16-40 ml of pure saline at 2-4 ml/s. A late phase scan was added 6 min after the injection of iodine contrast media using the prospective ECG-gating technique targeting the diastolic phase when intracardiac thrombus was suspected in the early phase images [20]. CT scan was performed with a slice thickness and tube voltage of 0.5 mm and 80-120 kV for 320-slice CT and 0.625 mm and 70 kV for 256-slice CT, respectively; the same tube voltage as for non-contrast scan was applied in all cases. Tube current was determined based on the auto exposure control system in the initial 33 patients, and almost maximum tube current was applied for the remaining 37 patients.

Analysis of ECV on CT
Myocardial ECV of the LVM was measured using commercially available software (Ziostation 2, Ziosoft Inc, Japan) with the following equation: ECV = (ΔHUm/ ΔHUb)/(1 − Hct), where ΔHUm is change in myocardial CT attenuation in Hounsfield units (HU), ΔHUb is change in CT attenuation of the blood, and Hct is hematocrit [20] ( Fig. 1). This software performs automatic three-dimensional non-rigid registration of the myocardium between non-contrast and late phase CT images to generate subtraction images [20]. The change in CT attenuation (ΔHU) is then obtained on the subtraction image. The software produces a polar map showing both the 16 American Heart Association myocardial segments with the mean ECV value for each segment and the mean ECV value of all LVM. The ECV of LVM was measured by a cardiologist with 4 years' experience in cardiac CT analysis (SY), and the ECV values were used for the succeeding analysis. ECV of LVM on CT was also measured by another cardiologist with 14 years' experience in cardiac CT analysis (HT), and inter-observer agreement was analyzed.
In addition, quantitative image quality analysis was performed by a manual drawing of about 10-mm 2 regions of interest (ROI) on the LE and about 50-mm 2 ROI on remote myocardium without LE. ROIs at the LE and the remote normal myocardium were manually drawn, and CT attenuation values were measured in the patients with LE of LVM on CT (Fig. 2). Contrast-to-noise ratio (CNR), defined as the difference in attenuation of LE and remote normal myocardium divided by the standard deviation (SD) of the remote normal myocardial attenuation, was also quantified [20,21]. CNR was compared between patients with a higher and lower body mass index (BMI). Both cardiac MRI and CT were performed in 27 of 70 patients (39%) within 3 months, and the diagnostic accuracy of CT for the detection of LE of LVM on the patient-based analysis was evaluated. CNR analysis was performed by a cardiologist (SY). The pattern of LE of LVM on CT was classified into septum mid, epicardium, endocardium, and overlap patterns. The effective dose for scanning of coronary arteries was calculated from the dose-length product in the dose report (conversion factor 0.014) [20].

Statistical analysis
Continuous variables are expressed as the mean ± SD or as median (interquartile range) and categorical variables as counts and percentages. Continuous variables were compared using Student's t-test, and the Fisher's exact test was used to compare categorical variables. The correlation coefficient was measured to evaluate consistency between the two observers' assessments of ECV of LVM on CT. All tests were 2-sided, and P values < 0.05 were considered to indicate statistical significance. The receiver operating characteristics (ROC) curve analysis was performed based on MACE. Best cut-off in ROC curve analysis was based on the Youden index. Kaplan-Meier analysis was used to calculate the time to MACE and to estimate MACE rates, and the log-rank test was applied for between-group comparisons. A Cox proportional hazards model was also performed to investigate the association between the time to MACE and predictor variables. All statistical analyses were performed using the JMP software program, version 15.0.0 (SAS Institute Inc, Cary, NC, USA).

Results
Patients were followed for 51 ± 47 months after cardiac CT. Twenty-one patients (30%) experienced MACE during the follow-up period. The breakdown of MACE in these 21 patients was one cardiac death, three ventricular tachycardia or fibrillation, and 17 admissions due to congestive heart failure. The number of patients taking mineralocorticoid receptor antagonists (MRA) was significantly higher in cases with MACE than in those without MACE (67% vs. 29%, P = 0.0039) ( Table 1). ECV of the LVM on CT was significantly higher in cases with MACE than in those without MACE (37.16 ± 5.91 vs. 32.59 ± 3.95%, P = 0.0031). The presence of significant valvular disease (≥ moderate) was significantly higher in cases with MACE than in those without MACE (57% vs. 20%, P = 0.0044). LVEF was significantly lower in cases with MACE than in those without MACE (23 ± 8 vs. 31 ± 11%, P = 0.0024). LV endo-systolic volume (LVESV) was significantly larger in  vs. 138 ± 78, P = 0.0452) ( Table 2). The best cut-off value of ECV on LVM for prediction of MACE was 32.26% based on receiver operating characteristics (ROC) analysis. The area under the curve (AUC) of the ROC curve was 0.74 (P = 0.0004) (Fig. 3a). Sensitivity and specificity for the prediction of future MACE were 95% and 51%, respectively (Fig. 3a). The best cut-off value for left ventricular ejection fraction for prediction of MACE was 24% based on ROC analysis. The AUC of the ROC curve was 0.70 (P = 0.0049). Sensitivity and specificity for prediction of future MACE were 71% and 67%, respectively. (Fig. 3b) The best cut-off value for LV endo-systolic volume (LVESV) for prediction of MACE was 107 ml based on ROC analysis. The area under the curve of the receiver operating characteristics curve was 0.66 (P = 0.022). Sensitivity and specificity for prediction of future MACE were 90% and 37%, respectively (Fig. 3c).
Cases with ECV ≥ 32.26% had significantly higher MACE than those with < 32.26% during the follow-up period based on Kaplan-Meier analysis (P = 0.0032) (Fig. 4). A univariate Cox proportional hazards model showed that significant valvular disease (≥ moderate) on TTE, LVEF on TTE ≤ 24% and ECV of LVM on CT ≥ 32.26% were significant risk factors for MACE (P = 0.0114, 0.0152 and 0.0150) ( Table 3). The multivariate Cox proportional hazards model showed that ECV of LVM on CT ≥ 32.26% was the only independent predictor of MACE during the follow-up period (P = 0.0354) ( Table 4). The correlation coefficient between ECVs of LVM on CT, which two observers measured, was 0.84.
CNR of 16 patients with LE of LVM on CT was 4.3 ± 1.2. CNR was not significantly different between the patients with higher BMI (> 21.26, the median value) than the others (3.8 ± 1.0 vs. 4.7 ± 1.3, P = 0.1441). LE was detected in 10 cases on MRI and eight on CT, giving a sensitivity, specificity, and overall accuracy of CT as compared with the reference standard, MRI, of 80%, 100%, and 93%, respectively.
The effective radiation dose for additional late phase scan was 3.5 ± 0.9 mSv (radiation dose for the first 7 cases was not recorded, and these were excluded from this analysis).

Discussion
The results of this study suggest that ECV of LVM on CT might be a predictor of future MACE in patients with DCM. CT is useful for the detection of coronary artery stenosis, and ECV analysis is also feasible if an additional late phase scan is performed. From these findings, CT appears to be a useful modality for whole cardiac screening in patients with DCM [6].

ECV analysis in cases on DCM
The presence of LGE is a marker of higher risk of future cardiac events in patients with DCM, because myocardial damage is a marker of low cardiac function and fatal arrhythmia [5,6]. However, almost two-thirds of patients with DCM do not have LGE, because the evaluation of LGE is a qualitative analysis of focal fibrosis, and diffuse myocardial fibrosis is difficult to detect as LGE. Recently, ECV measurement using T1 mapping on MRI has become available for the prediction of future MACE [7]. MRI is sometimes contraindicated in cases with DCM because of implanted mechanical devices or claustrophobia, and gadolinium contrast is also contraindicated in cases with renal dysfunction [9]. Even when MRI is possible, image quality often deteriorates, and acquisition time is usually longer in DCM cases with arrhythmia, e.g., atrial fibrillation or difficulty in breath-holding due to chronic heart failure. In these situations, the specialized skills of radiology technologists are necessary to obtain higher-quality images. Additionally, because T1 mapping is a new sequence of cardiac MRI, performance is limited to the latest scanners in well-equipped institutions. Cardiac CT, including late-phase scans, is easily performed in a few minutes without difficulties even in cases with arrhythmia or difficulty in holding breath, because each scan time is short. Additionally, CT is more accessible than MRI because the number of CT Fig. 3 Receiver operating characteristics analysis for prediction of major adverse cardiac events. The best cut-off value for extracellular volume fraction (ECV) on the left ventricular myocardium for prediction of major adverse cardiac events (MACE) was 32.26% based on receiver operating characteristics (ROC) analysis (a). The area under the curve of the receiver operating characteristics curve was 0.74 (P = 0.0004) (a). The best cut-off value for left ventricular ejection fraction for prediction of MACE was 24% based on ROC analysis (b). The area under the curve of the receiver operating characteristics curve was 0.70 (P = 0.0049) (b). The best cut-off value for endosystolic volume for prediction of MACE was 107 ml based on ROC analysis (c). The area under the curve of the receiver operating characteristics curve was 0.66 (P = 0.022) (c) machines is almost twice the number of MRIs in Japan [22]. Image quality of LE on CT may be lower in patients with a higher BMI, but CNR did not significantly differ between eight patients with higher BMI than the other eight in 16 patients with LE on CT.
Increased ECV on LVM is mostly consistent with the higher amount of biopsy-proven myocardial fibrosis [23]; accordingly, higher ECV means severe degeneration of LVM, which leads to lower LV function or ventricular arrhythmic events [8]. The amount of myocardial fibrosis is a significant risk factor of future cardiac events in several myocardial diseases [5,6,8].

Factors related with the patient prognosis
Several clinical factors were suspected as risk factors in cases with DCM in this study, including ECV of LVM on CT, LVEF, LVEDV, LVESV on TTE, and significant valvular disease. However, only ECV of LVM on CT was an independent risk factor of MACE in cases with DCM. The results of TTE, including LVEF, LVEDV, and LVESV, were affected by the status of the control of chronic heart failure, presence of arrhythmia, and the physique of the patients. The degree of valvular regurgitation or stenosis on TEE were also affected by the several conditions at the time of TTE.
On the other hand, ECV of LVM reflects the amount of myocardial fibrosis, as mentioned above, and the value of ECV of LVM is accordingly more stable than the parameters of TTE. ECV of LVM is increased in patients with hypertensive heart disease (HHD), especially in those with both HHD and lower LVEF [24]. ECV of LVM is also increased in the patients with atrial fibrillation, and is higher in those with persistent AF than in those with paroxysmal AF [25]. ECV of LVM is also well correlated with the degree of mitral valve regurgitation [26]. ECV of LVM is therefore the most sensitive parameter, and is a reflection of the presence of complex cardiac abnormalities, including LV function, LV afterload, arrhythmia and valvular disease. Additionally, the ECV of LVM starts to increase in cases in the pre-disease stage of DCM, a phenomenon well expressed by the fact that ECV on LVM is more sensitive than LVEF [2].

Additional radiation dose for late enhancement analysis on CT
ECV analysis on CT requires an additional radiation dose, but this is a tradeoff for the important clinical information obtained. The effective dose for the late-phase scan in this study was 3.5 ± 0.9 mSv, which is smaller than the effective dose for chest CT for evaluation of the lung (almost 5 mSv) [27]. The radiation dose for late phase cardiac images has recently decreased following the introduction of new iterative reconstruction techniques and wide coverage

Improvement in image quality of late enhancement
Contrast resolution of late phase cardiac images on CT is inferior to MRI, and MRI remains the gold standard for evaluation of myocardial fibrosis on late phase images. Although the CT attenuation value increases when a CT scan is performed using a lower tube voltage, image noise increases owing to the limited radiation dose [28]. Recently, however, newer iterative reconstruction techniques have appeared, and the maximum tube current of CT scanners has increased. These help decrease image noise in images acquired using a lower tube voltage. We previously reported that the combination of new-generation CT and the iterative reconstruction technique is useful for improving the image quality of late enhancement on CT, and for higher diagnostic accuracy in the detection of myocardial fibrosis [12]. Of note, the improvement in image quality in late enhancement of LVM has resulted in the recommendation of ECV analysis using CT as a substitute for MRI in the latest guidelines for cardiac CT and cardiac amyloidosis from the Japanese Cardiovascular Society [29,30]. The presence of LE of LVM on CT did not differ between patients with and without MACE. This result is not consistent with a previous report on MRI [5]. We considered that this was attributable to the lower image quality of LE on CT than on MRI. Visual assessment of LE of LVM on CT is sometimes hampered by lower image quality. Nevertheless, ECV analysis is a suitable alternative, and we clarified that this analysis could provide additional clinically significant information in DCM. CNR of 16 patients with LE of LVM on CT was 4.3 ± 1.2. CNR of LE on CT in cases with old myocardial infarction were close to 6 and 2 in the two previous studies; the CNR value of LE in the present study was not particularly different from that in the previous studies of old myocardial infarction even though the background myocardial disease was different [20,21].

Limitations
Several limitations of our study warrant mention. First, the study was conducted under a retrospective design at a single center. Second, ECV analysis was performed on singleenergy images, and subtraction of late phase and non-contrast images was therefore necessary. Gaps between images in these two phases might cause under-or overestimation of ECV on single-energy images compared with analysis of dual-energy images using the latest CT scanners (without gaps). Finally, the study was conducted using different CT scanners, and differences between them may have affected the results.

Conclusion
Evaluation of ECV of LVM on CT helps predict MACE in patients with DCM.