Utilization of Synthetic Hematocrit Derived from Cardiac MRI for Estimating Extracellular Volume

Objective: For a correct estimation of extracellular volume, patients’ blood samples must be taken at least within 24 hours after cardiac magnetic resonance imaging, which may cause problems in immediate or retrospective evaluation. The aim of this study is to demonstrate the applicability of synthetic hematocrit for the calculation of extracellular volume. Materials and Methods: Cardiac magnetic resonance images including T1 mapping were retrospectively reviewed. Medical records of each patient were also collected and patients who had blood samples including hematocrit values within 24 hours of cardiac magnetic resonance imaging were included in the study. Simple linear regression along with intraclass correlation coefficient were used to calculate the coefficient of determination and agreement between the two measurements of hematocrit and extracellular volume values. The compatibility of the synthetic extracellular volume with the native extracellular volume was analysed using “Kappa Analysis”. Results: Synthetic and native hematocrit values showed significant correlation (p<0.001). Synthetic hematocrit had a high intraclass correlation coefficient (0.833). Kappa analysis for each region (Apex κ=0.88, Mid κ=0.75, Basal κ=0.87) showed significant interobserver reliability (p<0.001). Conclusion: Synthetic hematocrit formulas for each vendor may help improve synthetic extracellular volume values without contrast administration.


A B S T R A C T
Objective: For a correct estimation of extracellular volume, patients' blood samples must be taken at least within 24 hours after cardiac magnetic resonance imaging, which may cause problems in immediate or retrospective evaluation. The aim of this study is to demonstrate the applicability of synthetic hematocrit for the calculation of extracellular volume.
Materials and Methods: Cardiac magnetic resonance images including T1 mapping were retrospectively reviewed. Medical records of each patient were also collected and patients who had blood samples including hematocrit values within 24 hours of cardiac magnetic resonance imaging were included in the study. Simple linear regression along with intraclass correlation coefficient were used to calculate the coefficient of determination and agreement between the two measurements of hematocrit and extracellular volume values. The compatibility of the synthetic extracellular volume with the native extracellular volume was analysed using "Kappa Analysis".

INTRODUCTION
Cardiac Magnetic Resonance Imaging (CMR), in addition to assessing cardiac anatomy and myocardial motion, is a part of the main work-up to study myocardial pathology due to its superior tissue characterization and enhancement properties. Normal myocardium improves homogeneously and is lately washed out. However, myocardial scar tissue, fibrosis, or edema show late gadolinium enhancement due to retention of contrast in the extracellular space, which has an increased volume in these pathologies [1]. Extracellular volume (ECV) indirectly indicates cardiac remodelling and is a quantitative parameter for prognosis of myocardial pathologies [2].
T1 mapping techniques grant quantification of fibrosis, but also allow the diagnosis of myocardial disease [3][4][5]. T1 mapping allows the estimation of the longitudinal relaxation time (known as the T1 value), i.e., the rate at which protons to regain their longitudinal magnetization after a radiofrequency pulse. T1 mapping can be obtained via balanced steady-state free precession (SSFP) based shortened or non-shorthened modified lock-locker inversion recovery (MOLLI) [5].
The evaluation of T1 values enables the ECV estimation. To calculate the ECV, the T1 values and the hematocrit value must be known, as shown in the formula [3]: For correct estimation of ECV, patients blood samples must be taken at least within 24 hours of CMR [3], which may cause problems with immediate or retrospective evaluation of the patient. Therefore, synthetic hematocrit estimation using longitudinal relaxation rate of blood without blood sampling was proved to be feasible [6]. The longitudinal relaxation rate (R1) is the frequency function of the T1 value (R1=1/T1) and the R1 value of blood has a linear relationship with the hematocrit [7,8].
Hematocrit which is derived from precontrast T1 blood value is called synthetic hematocrit [6]: This study was designed to test the applicability of synthetic hematocrit measurement in our own patient population by different observers.

MATERIALS and METHODS
Approval for the study was obtained from Instutional Ethics Board. Informed consent was not obtained as this was a retrospective study.
Between November 2017 -July 2018, CMR including T1 mapping were retrospectively reviewed. Medical records of each patient were also collected and patients who had blood samples including hematocrit values within 24 hours of CMR imaging were included in the study.
CMR was performed on a 1.5T scanner (Aera 1.5T, Siemens Healtcare, Germany). T1 mapping sequences were acquired using balanced SSFPbased MOLLI for basal, middle, and apical short-axis slices of the left ventricle. Imaging parameters for each patient were: 358x420 mm field of view, 1. Image analysis was performed using Syngo.via software (Siemens Healtcare, Germany). Region of interest (ROI) was manually drawn by a radiologist in basal, middle, and apical slices of the left ventricle to obtain both native and postcontrast T1 values of the myocardium. Epicardial fat and blood pool were avoided in ROI placement ( Figure 1). Another ROI was drawn at the blood pool avoiding the papillary muscles to obtain native and postcontrast T1 values of the blood.
Synthetic hematocrit values were estimated using the above formula. Both native and synthetic ECV values were also estimated using the native and synthetic hematocrit values, respectively.
Statistical analysis was performed using IBM SPSS Statistics 23 software. Simple linear regression was used to calculate the coefficient of determination and "Intraclass Correlation Coefficient" was used to evaluate the agreement between both measurements of hematocrit and ECV values. The compatibility of the synthetic ECV with the native ECV was analysed using "Kappa Analysis" after classifying both measurements with a cut-off value of 28%.

RESULTS
Sixty-one patients; 33 male and 28 female, fulfilled the aforementioned criteria and were included in the study. Indications for CMR imaging were non-ischemic cardiomyopathy (n=48), myocardial infarction (n=7), myocarditis (n=5) and constrictive pericarditis (n=1). The mean age was 52 years (range 11-80 years), and 7 patients were under 18 years of age.
Mean native hematocrit was 40.4%, while hemoglobin levels ranged from 12.1 gr/dl -15.8 gr/dl. The mean synthetic hematocrit value was 40.1%. Also, the mean synthetic ECV values for each segment were 29.2% for basal segment, 28.1% for mid segment and 27.7% for apex segment while native values were 29.1%, 28.3% and 28.1% respectively (Table 1).

DISCUSSION
Our results showed that synthetic ECV values estimated from three different regions of the myocardium showed excellent correlation with native ECV.
ECV represents the extracellular matrix and it is known to increase in various pathologies such as edema and fibrosis. The ECV value has an impact on both diagnosis and prognosis, which is why its evaluation has become a routine part of CMR examination.
Similar to our study, a couple of studies using both 1.5T and 3T magnets found a significant correlation between native and synthetic hematocrit and ECV values [9,10]. Kammerlander et al. proposed their own synthetic hematocrit formula for their vendor, which had a stronger correlation compared to the previously published one [6].
On the other hand, two studies found moderate [11,12] and one study found poor [13] However, the ECV values in these studies showed excellent correlation, which was explained by the formula, since only significant changes in the hematocrit value can have an effect on the ECV, while the other parameters remain constant. Although we found an excellent correlation between native and synthetic hematocrit, the correlation was stronger for ECV, which may be the result of this calculation.  [11,13]. Further studies involving different regions of the myocardium may also suggest localization for accurate measurements of synthetic ECV.
Synthetic ECV values have a significant correlation with native values, but large-scale studies with measurements from different regions of the myocardium need to be performed to find more accurate results. Specific hematocrit formulas for each manufacturer could also help to improve synthetic ECV values.

CONFLICT of INTEREST STATEMENT
The authors have no conflict of interest.

Figure2.
The graphs, acquired via simple regression analysis. The proximity of each dots shows the high validity. Note that the validity for synthetic estimation is higher on ECV calculation.