Comparison of Left Ventricular Electromechanical Mapping and Left Ventricular Angiography

Abstract

We performed this prospective cohort study to correlate the findings of left ventricular angiography (LVA) and NOGA™ left ventricular electromechanical mapping (LVEM) in the evaluation of cardiac wall motion and also to establish standards for wall motion assessment by LVEM. Fifty-five patients (35 men; mean age, 60.4 ± 11.8 years) eligible for elective left cardiac catheterization underwent LVA and LVEM. Wall motion scores, LV ejection fractions (LVEF), and LV volumes derived from LVA versus LVEM data were compared and analyzed statistically. Receiver operating characteristic (ROC) curves were used to assess the accuracy of LVEM in distinguishing between normal, hypokinetic, and akinetic/dyskinetic wall motion. Mean LVEM procedure time was 37 ± 11 minutes. The LVEM and LVA findings differed for mean LVEF (55% ± 13% vs 36% ± 9%), mean end-systolic volume (56 ± 13 mL vs 36 ± 10 mL), and mean end-diastolic volume (174 ± 104 mL vs 123 ± 65 mL). Mean wall motion scores (± SD) for normokinetic, hypokinetic, and akinetic/dyskinetic segments were 13.9% ± 5.6%, 8.3% ± 5.2%, and 3.2% ± 3.1%, respectively. Cutpoints for differentiating between wall motion types were 12% and 6%. The ROC curves showed LVEM to have a 93% accuracy in differentiating between normokinetic and akinetic/dyskinetic segments and a 73% accuracy between normokinetic and hypokinetic segments. These data suggest that LVEM can differentiate between normal and abnormal cardiac wall motion, although it is more accurate at differentiating between normokinetic and akinetic/dyskinetic motion than between normokinetic and hypokinetic motion. (Tex Heart Inst J 2003;30:19–26)

Left ventricular electromechanical mapping (LVEM) using NOGA™ software (Biosense-Webster; Diamond Bar, Calif) is a new technology that reconstructs 3-dimensional maps of the left ventricle (LV) from data acquired at multiple points on the endocardium. The NOGA software is used to compare the location of an endocardial point in systole and diastole and calculate its movement in relation to other surrounding points. This movement is expressed as linear local shortening (LLS), which is a validated measure of myocardial mechanical function. ^1–3 Through reconstruction of the LV endocardial contour, the system has the capability to provide hemodynamic data such as LV ejection fraction, end-systolic volume, and end-diastolic volume. ⁴

Left ventricular angiography (LVA) was the 1st method used to assess LV wall motion contractility and hemodynamic parameters. ^5,6 Other methods used for this purpose now include 2- and 3-dimensional (2-D and 3-D) echocardiography, radionuclide ventriculography, computed tomography, and magnetic resonance imaging. ⁷ Left ventricular angiography is still widely used for LV assessment and remains one of the gold standards for wall motion analysis.

Previous studies have demonstrated a moderate correlation between LVEM and LVA in terms of global and regional contractile LV function and volume measurements. ^8–10 To determine whether this correlation remains valid for LVEM findings obtained with use of a newer version of the NOGA software (v. 4.0), which incorporates a different LLS algorithm, we analyzed and correlated the findings of LVEM and LVA in a prospective cohort study. In addition, as the primary end point, we sought to establish standard values for wall motion assessment (in comparison with LVA findings) that can be used routinely in the analysis of LV electromechanical maps.

Patients and Methods

Study Design and Inclusion Criteria

We conducted a cohort prospective study of 55 patients who underwent mapping procedures after elective left cardiac catheterization at 2 centers (Texas Heart Institute, Houston, Texas; and Hospital Pro-Cardíaco, Rio de Janeiro, Brazil). Electromechanical mapping was performed after LVA. The LVEM procedures were performed only in patients who were clinically stable; excluded were those patients who had severe peripheral vascular disease, atrial fibrillation, aortic stenosis, suspected thrombus in the left ventricle, or acute myocardial infarction. The study protocol was approved by the ethics committees of both hospitals. There was no industry support for this study. The procedures were explained, and informed written consent was obtained from all patients before they were enrolled in the study.

NOGA Mapping System and Technique

A NOGA electromechanical map of the left ventricle is constructed by the acquisition of a series of points at multiple locations on the LV endocardial surface gated to a surface electrocardiogram. NOGA LVEM uses ultra-low magnetic fields (10⁻–10⁻⁶ tesla) that are generated by a triangular magnetic pad positioned beneath the patient. The intersection of the magnetic fields with a location sensor just proximal to the deflectable tip of a 7F mapping catheter helps to determine the location and orientation of the catheter tip inside the left ventricle. An algorithm is used by the NOGA system to calculate and analyze the movement of the catheter tip or the location of an endocardial point throughout systole and diastole. That movement is then compared with the movement of neighboring points in an area of interest. The resulting value, which is called linear local shortening (LLS) and is expressed as a percentage, is representative of the mechanical function of the left ventricle at that point.

Data points are obtained only when the catheter tip is in stable contact with the endocardium. This contact is determined automatically by the NOGA system using the following criteria: 1) point loop stability (LS), defined as the trajectory of a specific point during 2 consecutive cardiac cycles (a low value, indicating good-quality data, is preferable); 2) cycle length (CL) stability, defined as the difference between the CL of a specific point and the average CL of the previously recorded 100 beats; 3) local activation time (LAT) stability, defined as the difference between the LAT of a point and the LATs of points previously recorded (variation should be no more than 3 ms); and 4) location stability (LcS), defined as the variability in the location of the catheter tip during the cardiac cycle (between end systole and end diastole). The mapping catheter also incorporates electrodes that measure endocardial electrical signals (unipolar or bipolar voltage). Voltage values are assigned to each point acquired during mapping of the LV, and an electrical map is constructed. Each data point has an LLS value and a voltage value. When the map is complete, all the data points are integrated by the NOGA workstation and are presented in a 3-D color-coded reconstruction of the endocardial surface, and in 9- and 12-segment bull's-eye views that show average values for LLS and voltage data in each segment (Fig. 1). These maps can be spatially manipulated in real time on a Silicon Graphics workstation (Mountain View, Calif).

Fig. 1 A representative, 3-dimensional, color-coded reconstruction of the endocardial surface (left) and 9- and 12-segment bull's-eye views (right top and bottom, respectively).

LLS = linear local shortening

The 3-D representations acquired during the cardiac cycle are used to calculate LV volumes. The NOGA system uses the largest volume as the end-diastolic volume (EDV) and the smallest volume as the end-systolic volume (ESV). The LV ejection fraction (LVEF) is calculated as (EDV – ESV)/EDV. The NOGA system also uses, as predetermined by the operator, a triangle fill threshold (FT) that determines the extent to which the computer algorithm will interpolate data into (or “fill in”) the space between adjacent points. In the present study, a triangle FT of 15 mm was used (the standard recommended by the manufacturer to ensure map completeness). After the acquisition of points, postprocessing analysis is performed with a series of filters in the moderate setting to eliminate inner points, points that do not fit the standard stability criteria (LcS <4 mm; LS <6 mm; and CL <10%), points acquired during ST-segment elevation, and points not related to the left ventricle (for example, those on the atrium).

Left Ventricular Angiography Protocol

Left ventricular angiography was performed through the femoral approach using a 5F pigtail catheter (Cordis; Miami Lakes, Fla), which was marked at 1-cm intervals for accurate calculation of the LV volumes and LVEFs according to the area-length formula. ¹¹ All LVAs were obtained in 2 planes—a 30° right anterior oblique (RAO) view and a 60° left anterior oblique (LAO) view—during a period of stable sinus rhythm. Ventricular volume was not measured during or after a premature beat. Wall motion was evaluated in the RAO view by 2 independent experienced observers. Segments visible in the angiographic RAO view (that is, anterobasal, anteromedial, apical, inferomedial, and posterobasal) were compared with the corresponding 5 segments of the LVEM bull's-eye view. Wall motion in each myocardial segment was scored as follows: 0 = dyskinetic; 1 = akinetic; 2 = hypokinetic; and 3 = normokinetic.

Left Ventricular Electro-mechanical Mapping Protocol

Left ventricular electromechanical mapping was performed as follows. All patients were heparinized (70 U/kg) after biplane LVA and before LVEM. The mapping catheter curve (B, D, or F) was selected on the basis of LV size. The catheter (7F) was advanced under fluoroscopic guidance to the descending thoracic aorta, where its tip was fully deflected and then subsequently advanced around the aortic arch and across the aortic valve into the left ventricle. Inside the left ventricle, the deflection was relaxed and the catheter tip was oriented toward the LV apex. The initial data point was acquired at the LV apex, and 2 points were acquired—one at the base of the septum and one at the lateral wall—to complete an initial triangle defining the borders of the LV. Subsequent points were acquired until all endocardial segments were uniformly sampled (ideally, 3 points in each of 12 segments, according to the NOGA™ Mapping Excellence Program^*). Each data point was filtered on-line, immediately after acquisition, and during postprocessing analysis with use of the parameters described above.

Map Analysis

The LVEMs were analyzed using data on the mechanical function of the myocardium (LLS), which has been extensively validated in previous studies. ^1–3 The data from the maps were compared with data obtained by angiography. Angiography provides a 2-D image of the left cardiac chamber that historically has been compared with images produced by emerging technologies such as echocardiography, radionuclide studies, and MRI. ⁷

To compare wall motion as represented by LVEM versus LVA, the anterobasal, anteromedial, apical, inferomedial, and posterobasal segments on the RAO angiogram (Fig. 2) were matched with those on the corresponding 5 segments of the 9-segment LVEM bull's-eye view. (Again, according to the NOGA™ Mapping Excellence Program, each LVEM segment should contain at least 3 points after initial filtering and postprocessing analysis.) The segments were divided into 3 groups according to wall motion score: Group I (0 or 1, akinetic or dyskinetic); Group II (2, hypokinetic); and Group III (3, normokinetic).

Fig. 2 Schematic illustrations show the 5 myocardial segments evaluated by left ventricular angiography and left ventricular electromechanical mapping (top) and corresponding bull's-eye view (bottom) for comparison.

AB = anterobasal; AM = anteromedial; Ap = apical; IM = inferomedial; PB = posterobasal

Statistical Analysis

Left ventricular ejection fraction and volume data were presented as the mean ± standard deviation (SD). The mean values obtained by LVEM and LVA were compared using the unpaired Student's t-test. A P value of <0.5 was considered significant for all comparisons. Correlation coefficients were reported in terms of the Pearson correlation index (r), and correction indexes for each variable were created. The Bland-Altman technique was used to determine the agreement between continuous measurements acquired by LVEM and LVA. For each group of wall motion scores, the mean LLS (± SD) was determined. Boundaries for classification of wall motion as normal, hypokinetic, and akinetic/dyskinetic were derived by discriminant analysis. This analytic technique finds the 2 boundaries that predict classification into a group (in this case, 1 of the 3 groups determined by LVEM wall motion score). The Scheffé test was also used for multiple comparisons of wall motion. Receiver operating characteristic (ROC) curves were used to assess the sensitivity and specificity of LVEM for distinguishing between normokinetic and hypokinetic and between normokinetic and akinetic/dyskinetic wall motion.

Results

The study population consisted of 55 consecutive patients who were scheduled for elective left heart catheterization, during which LVEM was performed (Table I). The mean age was 60.4 ± 11.8 years (range, 42–86 years), and there were more men (35 [64%]) than women. The average LVEM procedure time was 37 ± 11 minutes. After application of a moderate automatic filter, a mean of 62 ± 12 points was obtained in each mapping to render a representative 3-D reconstruction of the LV. There were no deaths or major complications associated with LVEM (such as malignant ventricular arrhythmias or LV perforation requiring pericardiocentesis). One patient had a left ventricular perforation that resulted in a small post-procedural pericardial effusion. This complication was managed conservatively and was followed with serial echocardiography until it resolved; pericardiocentesis was not necessary.

TABLE I. Characteristics of 55 Patients at Baseline

The mean values for EF, EDV, and ESV, obtained by LVEM and LVA, are shown in Table II. The Pearson correlation index and graphs for each variable are shown in Figs. 3–5. A moderate-to-good correlation was found between LVEM and LVA findings; however, as shown by Bland-Altman analysis (Figs. 6–8), clinical disagreement and lack of interchangeability was found between all measured parameters for both methods.

TABLE II. Mean Values for LVA and LVEM Findings

Fig. 3 Pearson correlation index for ejection fraction (EF) as determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

Fig. 4 Pearson correlation index for end-diastolic volume (EDV) as determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

Fig. 5 Pearson correlation index for end-systolic volume (ESV) as determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

Fig. 6 Bland-Altman association for ejection fraction (EF) as determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

Fig. 7 Bland-Altman association for end-diastolic volume as (EDV) determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

Fig. 8 Bland-Altman association for end-systolic volume (ESV) as determined by left ventricular angiography (LVA) and left ventricular electromechanical mapping (LVEM).

In the 55 cases studied, the wall motion scores in a total of 275 segments were determined by LVEM and LVA, compared, and classified into 3 groups. Group I (akinetic/dyskinetic) comprised 9 segments; Group II (hypokinetic), 68 segments; and Group III (normokinetic), 198 segments. Table III shows the mean LLS (± SD) values for each group. Using discriminant analysis, the cutpoints for differentiation between normokinetic, hypokinetic, and akinetic/dyskinetic segments by LLS were set at 12% and 6% (Table IV). As shown by ROC curves (Figs. 9 and 10), the LLS values determined by LVEM had an accuracy of 93% in differentiating between normokinetic and akinetic/dyskinetic myocardial segments and an accuracy of 73% in differentiating between normokinetic and hypokinetic myocardial segments.

TABLE III. Wall Motion Classification and Mean LLS Values

TABLE IV. Cutpoints for Differentiation of Wall Motion by LLS According to Discriminant Analysis

Fig. 9 Receiver operating characteristic curve for the differentiation between normokinetic and akinetic/dyskinetic segments by local linear shortening (LLS).

Area under the curve = 0.93; threshold <6%; sensitivity, 89%; specificity, 88%

Fig. 10 Receiver operating characteristic curve for the differentiation between normokinetic and hypokinetic segments by local linear shortening (LLS).

Area under the curve = 0.73; threshold <12%; sensitivity, 68%; specificity, 62%

Discussion

The present study was designed to compare the assessment of global and regional function of the left ventricle by 2 different techniques: LVEM and LVA. Both techniques are invasive and yield quantitative and qualitative information about the performance of the left ventricle. However, LVEM creates an on-line, real-time 3-D reconstruction of the endocardial surface and has proven value for assessing both myocardial viability ¹² and the mechanical function of myocardial segments.

Several studies have focused on the ability of LVEM to distinguish between normal and infarcted myocardium, to enable comparison of hemodynamic parameters, and to perform wall motion analysis. Nonetheless, no practical cutpoints for LVEM analysis of wall motion had previously been established. ^8–10 Therefore, as the primary end point of a study of the largest LVEM cohort in the literature, we have correlated the findings of LVEM and LVA by analyzing only those LVEM segments that were also well visualized by LVA and well matched to the same regions of the LVEM bull's-eye views.

The values used to define wall motion in the present study were similar to those already described in the literature. However, the mean values (± SD) that we established for normokinetic, hypokinetic, and akinetic/dyskinetic wall motion (13.9% ± 5.6%, 8.3% ± 5.2%, and 3.2% ± 3.1%, respectively) overlapped and sometimes made it difficult to differentiate between normokinetic and hypokinetic tissue. This was well exemplified by the ROC curve (Fig. 10), which showed a weak accuracy of 73%. These findings agree with the data of Lessick and colleagues, ¹³ who compared echocardiography with LVEM data and reported an accuracy of 69%. On the other hand, our cutpoints allowed excellent differentiation between normokinetic and akinetic/dyskinetic myocardial areas, with an accuracy of 93% (Fig. 9).

The discrepancy in accuracy of differentating between normokinetic and hypokinetic tissue and between normokinetic and akinetic/dyskinetic tissue may be due to several factors. First, it can be very difficult in some cases, especially in dilated or very hypertrophic ventricles, to perform a complete LVEM because the catheter cannot uniformly reach all areas of the left ventricle. As a result, mapping may be incomplete, and the average of the mapped values for a determined region may be misrepresented. Second, data suggesting low contractility of basal areas of the LV may in part represent the presence of fibrous tissue in perivalvular areas. Future development of LVEM in terms of performance and analysis may improve upon the completeness of mapping and make the interpretation of the data more accurate.

As shown in a recent study by Van Langenhove and coworkers, ¹⁴ the hemodynamic data obtained by LVEM has its limitations. In their study, the correlation between ESV and EF measurements obtained by LVEM versus LVA was moderate (r = 0.67 and r = 0.78, respectively), and the correlation between EDV measurements was poor (r = 0.40). ¹⁴ These results differ somewhat from ours, which showed better correlations for measurements of ESV (r = 0.81) and EDV (r = 0.71) and a worse correlation for measurements of EF (r = 0.56). Except for this difference, the results of our Bland-Altman analysis and the results from Van Langenhove's study were similar, demonstrating that the dispersion of values is great and that the real clinical application of EDV and EF measurements supplied by LVEM is questionable, even when correction formulas are applied. Therefore, we do not advocate the routine use of the NOGA method as an alternative to other established methods for assessing LVEF.

Similar problems with overlap of wall motion values obtained by LVEM have been found in other series. ^8–10,15 For example (although their comparison was between the findings of LVEM and nuclear perfusion studies), Kornowski's group ¹² recorded normal and abnormal LLS values (normal, 12.5% ± 2.8%; abnormal, 3.4% ± 3.4%), that were very similar to ours in the present study (normal, 13.9% ± 5.6%; abnormal, 3.2% ± 3.1%).

Despite its limitations, LVEM may assume an important role in new therapies that directly target ischemic heart disease. Such therapies aim to promote angiogenesis or restore contractility through the transplantation of stem cells ¹⁶ and myoblasts ^17,18 or by injection of growth factors. ¹⁹ The future success of these and similar therapies depend profoundly on carefully controlled clinical trials that apply appropriate methods and end points. In that regard, the NOGA mapping system has an advantage over other potential therapeutic delivery systems (especially those involving surgery), because it is less invasive. In theory, the demarcation of appropriate “target” zones for treatment is one of the keys to the success of our procedure. Therefore, by defining practical LLS thresholds for assessing mechanical activity in the present study, we believe we have made it easier to target viable myocardium (that is, tissues with low LLS and preserved electrical activity) using the NOGA system and so optimize therapy. Moreover, even though the system is somewhat limited in its ability to differentiate wall motion, this limitation appears to be restricted to severely dilated or very hypertrophic ventricles. In such cases, performing a complete LVEM is already technically challenging. We believe that this limitation can be overcome by devising mapping procedures that are more careful and detailed.

In conclusion, our data indicate that there is a moderate correlation between LVA and LVEM findings and that LVEM can differentiate between normal and abnormal cardiac wall motion. However, it appears that LVEM is severely limited in its ability to measure LV hemodynamics, which will limit its widespread use for this purpose. Nevertheless, our findings are important, because they add to the current knowledge about the interpretation of LVEM findings, and because they have important implications for the use of LVEM in conjunction with intramyocardial therapies in which optimal treatment delivery requires the accurate targeting of viable (normokinetic) versus nonviable (akinetic/dyskinetic) segments.