Study population
This case-control, single center study was approved by the institutional ethics committee. A total of 150 patients with systolic dysfunction (30 consecutive cases with thrombus and 120 consecutive controls without thrombus) who underwent cine, adenosine stress, and LGE-CMR during the same period at Siriraj Hospital (Mahidol University, Bangkok, Thailand) between August 2014 and January 2016 were enrolled. Systolic dysfunction was defined as LVEF by cine-CMR below 40%. Patients were most commonly referred for CMR examination to assess myocardial ischemia, myocardial viability, or causes of cardiomyopathy. Patients exhibiting the following criteria were excluded: 1) inability to perform CMR due to a permanent pacemaker or implantable cardioverter defibrillator implantation, 2) incomplete CMR examination, 3) history of allergy to gadolinium, 4) pregnancy, and (5) history of claustrophobia.
On the day of the CMR procedure, a complete medical history including cardiac risk factors, current medications including antithrombotic and heart failure drugs, information regarding coronary angiography results, prior coronary revascularization, previous myocardial infarction, and thromboembolic events were collected to assess potential predictors of thrombus.
Additionally, regarding the excellent ability of LGE-CMR for the diagnosis of infarcted myocardium, patients were considered to have ischemic cardiomyopathy if there was transmural or subendocardial scar pattern from LGE-CMR. All other patients were classified as having nonischemic cardiomyopathy.
CMR protocol
Cardiac structure, function and thrombus were assessed from two CMR techniques, cardiac function by a SSFP technique and LGE using a 1.5 Tesla Gyroscan NT Philips scanner (Philip Medical Systems, Best, The Netherlands). Functional study was performed by the acquisition of images by a SSFP technique in Short-axis images were acquired every 8 mm throughout the entire LV. Long-axis images were obtained in standard 2-, 3-, and four-chamber orientations. Parameters for cardiac function were as follows: repetition time/echo time/number of excitations = 3.7 ms/1.8 ms/ 2, 390 × 312 mm field of view, 256 × 240 matrix, 1.52 × 1.21 reconstruction pixel, 8 mm slice thickness, and 70 flip angles.
For LGE imaging, images were acquired 15 to 20 min after intravenous injection of 0.2 mmol/kg gadolinium contrast agent (Magnevist, Bayer Schering Pharma, Berlin, Germany) with the following scanning parameters; echo time 1.25 ms, repetition time 4.1 ms, 15-degree flip angle, 303 × 384 mm field of view, 240 × 256 matrix, in-plane resolution 1.26 × 1.5 mm, slice thickness 8 mm and 1.5 Sensitivity-Encoding (SENSE) factor.
Additionally, long inversion time sequence was applied for increasing the accuracy for detection of LV thrombus, as mentioned in previous study [5].
CMR analysis
CMR indices of LV function, geometry, and scarring were measured to determine whether these parameters were related to the presence of thrombus. CMR images in short-axis view were classified as the basal, mid, or apical part of the left ventricle. Segmentation of each slice was performed according to the recommendation of American Heart Association [8]. Left ventricular volume and mass was calculated and indexed for the body surface area. LVEF was quantitatively assessed by using end-systolic and end-diastolic volume calculated from the multiple slice short axis images. Wall motion of each myocardial segment was recorded as presence or absence of abnormal wall motion. Wall motion of each myocardial segment was also recorded as 5-grade system as follows: 1 = normal, 2 = hypokinesia, 3 = akinesia, 4 = dyskinesia, and 5 = aneurysm. LGE images were analyzed by visual assessment. The transmural extent of LGE was graded as follows: 0 = no LGE, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, and 4 = 76–100% compared to the myocardial area in that segment. Total scar volume as a percentage of LV myocardium was defined as the sum of all transmural extent of infarction scores throughout LV wall thickness / 4 times total number of segments [9]. Transmural scar was defined as transmural extension of LGE more than 50% of myocardial thickness in each segment, subendocardial scar was defined as transmural extension of LGE less than 51% of myocardial thickness.
LV thrombus assessment
Left ventricular thrombi detected by CMR on cine sequences were defined as filling defects within the LV cavity, typically adherent to regions of abnormal wall motion (hypokinesis, akinesis, or dyskinesis) then confirmed by LGE technique in Fig. 1a. In LGE images, thrombus was diagnosed as an LV mass with post-contrast inversion-recovery characteristics consistent with avascular tissue. Moreover, LGE-CMR can be further assessed for thrombus identification by increasing the inversion time (i.e., 600 msec) to selectively null avascular tissue such as thrombus. This ‘long inversion time’ approach provides an image that renders black thrombus and surrounding bright myocardium (Fig. 1b) [10].
Geometrical parameters calculation
Here, two indices were evaluated as sphericity index and ‘new’ apical area index. The sphericity index was described as the LV long-axis length divided by the mid LV diameter at end diastole for each in Fig. 2 [11, 12].
To represent the importance of apical remodeling, a ‘new’ apical area index was introduced to evaluate the predictor of thrombus. The apical area index was simply calculated as a percentage of an apical-half LV area (mid to apex) divided by an entire LV area from a cine four-chamber view in end diastole. Similar to the sphericity index, LV long-axis length was measured from the endocardial border of the mitral valve annulus plane to tip of apex, and the mid LV diameter was measured at the half-length from the LV long axis to tip of apex (Fig. 3).
Intra- and inter-observer reliability
To assess intra- and inter-observer reliability of apical area index measurements by CMR, 20 patients were randomly selected to measure variability by the same observer 4 weeks after the initial analysis, and by the second independent observer blinded to the initial results.
Statistical methods
Descriptive statistics, including frequency and percentage, were used for categorical variables. Continuous variables were reported as mean ± standard deviation for normally distributed variables and median (percentile 25 and percentile 75) for non-normally distributed variables. Normality of distribution of variables was examined by Kolmogorov-Smirnov test. Comparisons of categorical variables between patients with and without thrombus were performed using chi-square test or Fisher’s exact test. Continuous variables were compared using Student’s t-test or Mann-Whiney U test. Univariate and Multivariate analyses of thrombus predictions were evaluated using binary logistic regression analysis (Backward method) and presented as Odds ratio (OR) (95% confidence interval [CI]). Intra- and inter-observer reliability were analyzed by the intraclass correlation coefficient.
For all tests performed, a two-tailed p-value < 0.05 was statistically significant. PASW Statistic (SPSS) 18.0 (SPSS, Inc., Chicago, IL, USA) was used to perform all statistical analyses.