healthmedicinet

Patient selection

From 2003 to 2013, we prospectively recruited consecutive patients being referred
for CRT. All patients had systolic heart failure (EF???35 % by transthoracic echocardiography),
QRS duration??120 ms, and New York Heart Association functional class II or III symptoms
despite optimal medical therapy. Patients were enrolled only if they would be able
to follow up 6 months after the CRT procedure and if they had no known contraindications
to CMR. The Emory University institutional review board approved the study and all
patients gave written informed consent prior to enrollment.

Electrocardiogram classification

A favorable electrocardiogram (ECG) was defined as true LBBB morphology and QRS duration??150 ms.
True LBBB morphology was classified as a QS or rS complex in V1 and/or V2; monophasic
R wave in leads I, aVL, V5, and V6; and mid QRS notching or slurring in at least two
of the following leads: I, aVL, V1, V2, V5, or V6. Non-favorable ECGs were those that
demonstrated an atypical LBBB, an intraventricular conduction delay not satisfying
criteria for true LBBB, or a QRS duration??150 ms. Given that significant intraventricular
conduction delay may exist in the presence of right bundle branch block 14], 18], patients with bifascicular block patterns were included in the analysis, but those
with isolated right bundle branch blocks were excluded.

Transthoracic echocardiography

Patients underwent two-dimensional transthoracic echocardiography at baseline and
at 6 month follow-up. The echocardiographic studies were performed on a General Electric
Vivid 7 (Milwaukee, Wisconsin). LV end-systolic volume (ESV), end-diastolic volume,
and EF were assessed by Simpson’s modified biplane method of discs using the apical
four-chamber and apical two-chamber views. All echocardiograms were reviewed by a
board-certified reader blinded to baseline and follow-up status.

Cardiovascular magnetic resonance

CMR was performed on a 1.5 T Siemens Avanto scanner (Erlangen, Germany) with a 5-element
phased array coil and ECG triggering. Steady-state free precession (SSFP) short-axis
images were acquired parallel to the mitral valve plane to cover the entire length
of the LV (8 mm slices with no slice gap). Two-, three-, and four-chamber cine images
were also acquired. Late gadolinium enhancement (LGE) CMR was performed with a phase-sensitive
inversion recovery sequence 10–15 minutes after the administration of 0.1 mmol/kg
MultiHance (gadobenate dimeglumine; Bracco Diagnostics, Singen, Germany). LGE images
were acquired in the basal, mid, and apical short axis, as well as the two-, three-,
and four-chamber views. Significant scar was defined as enhancement in??15 % of LV
myocardium 19].

Left ventricular wall motion analysis

Endocardial borders were traced on each frame of the short-axis cine images and radial
displacement curves were generated as previously described 20]. Briefly, radial displacement curves were generated by measuring the radial distance
of the endocardial contour relative to the LV centroid at 100 circumferentially spaced
points for each slice. To account for translation of the LV over the cardiac cycle,
the LV centroid was determined from the location of the mitral valve annulus and apex
on every frame in the two and four-chamber views. Regional wall motion delay times
were determined by cross-correlating each radial displacement curve to a patient-specific
reference curve and recording the delay time for peak correlation. Regional radial
displacement curves were compared visually to long and short axis cines for regional
myocardial thickening and LGE images to determine akinetic segments with passive movement,
which were excluded from wall motion analysis.

Regional wall motion delays were determined throughout the LV (excluding the apex)
and then mapped to a modified American Heart Association 17-segment model 21] (Fig. 1). LV wall motion patterns were categorized as type I if the wave front proceeded homogenously from the septum to the LV free wall (no
adjacent early and late segments) and type II if the wave front was heterogeneous with evidence of an inferred line of block (adjacent
early and late segments; Fig. 2). Septal flash was identified by rapid inward and outward motion during isovolumic
contraction involving at least one of the septal segments. Isovolumic contraction
time was characterized as the interval from the onset of LV contraction to aortic
valve opening as visualized on long-axis cine SSFP images and confirmed by radial
displacement curve analysis. In areas of septal flash, the time to peak radial displacement
was defined as the initial radial displacement, rather than a subsequent peak likely
representing rebound.

Fig. 1. CMR processing and wall motion pattern analysis. Endocardial contours were traced
on short-axis cine images (a; red circle) and the distance to the centroid computed for each site (b). Each color curve represents a corresponding colored line from a; 100 sites were sampled per slice, however, only 12 are shown here for graphical
simplicity. Each regional radial displacement curve is compared by cross-correlation
(sliding in time) to a patient-specific reference (b; yellow dotted line) to determine the mechanical delay time. Delay times are mapped to a modified American
Heart Association 17-segment model (c). Note the early motion in the septal segments in b, shown in red in c, represents the septal flash

Fig. 2. Type I and type II wall motion patterns. Modified American Heart Association models demonstrating (a) a type I wall motion pattern, with dotted lines indicating homogenous wave fronts anteriorly and inferiorly towards the lateral wall,
and (b) a type II wall motion pattern, with an inferior line of block (green line) and dotted line indicating an anteriorly directed wave front and late inferior wall motion

Identification of the latest contracting site

The number of discrete myocardial sites sampled per segment ranged from 50–100, depending
on the number of short axis slices obtained (i.e. the LV length) and the apical versus
mid or basal position. The latest contracting site was defined as latest single site
to reach maximum radial displacement. The segment containing this site was defined
as the latest contracting segment.

Coronary venography and device implantation

The implanting electrophysiologist had no knowledge of the CMR derived wall motion
patterns prior to implantation. All CRT devices were implanted using standard procedure.
Briefly, central venous access was gained via either the subclavian vein or cephalic
vein cut-down. Biplane balloon catheter coronary venography defined coronary venous
anatomy in the 30° right-anterior-oblique and 30° left-anterior-oblique views. A suitable
anterolateral, lateral, or posterolateral coronary vein was selected based solely
on anatomic characteristics; the CMR data was not used to guide this decision.

Left ventricular lead localization

Final LV lead position was determined by biplane fluoroscopy and comparison to baseline
venous anatomy by coronary venography. The right anterior oblique image defined long-axis
position, while the left anterior oblique image defined short-axis position (Fig. 3). Using these images, LV lead position was mapped to the modified American Heart
Association 17-segment model. As described in previous studies 8], 9], final LV lead placement was considered concordant if placed in a viable segment within 1 segment of the latest contracting segment,
and remote if placed more than 1 segment from the latest contracting segment. If the LV lead
was placed in a non-viable segment with??50 % transmural LGE, lead position was considered
remote.

Fig. 3. Left ventricular lead localization. Biplane venograms (a; right anterior oblique 30°, b: left anterior oblique 30°) and lead localizing still frame (c; right anterior oblique 30°, left anterior oblique 30° not shown) to map left ventricular pacing lead locations
(green ellipse on c) onto the modified American Heart Association model (d; green circle denotes lead location). Left anterior oblique images were used to determine the circumferential
location while right anterior oblique images were used for longitudinal position

Study outcome

The primary outcome was positive echocardiographic response to CRT, defined as reverse
LV remodeling with a reduction in ESV by???15 % at 6 months. We compared response
in patients with both a type II pattern and concordant lead (T2CL) to those without both of these findings (i.e.
type I pattern and/or remote leads).

Statistical analysis

Statistical analysis was performed with SPSS version 20 (Chicago, Illinois) and SAS
version 9.3 (Cary, North Carolina). Dichotomous variables were expressed as numbers
and percentages, while continuous variables were expressed as mean?±?standard deviation.
Fisher’s Exact test was used to compare categorical variables. Continuous variables
were compared by the Student’s t-test (normal distribution) or the Mann–Whitney U test (non-normal distribution). Predictors of CRT response were analyzed by univariate
binary logistic regression analysis. Independent predictors were analyzed by multivariate
binary logistic regression analysis in a model including traditional variables associated
with CRT response (favorable ECG and non-ischemic cardiomyopathy, NICM). A 1000 sample
bootstrapping analysis was done to assess internal validity of the multivariate model
using SPSS. In an exploratory analysis, novel imaging variables (presence of significant
scar, maximum opposite wall delay, and septal flash on CMR) were added to the multivariate
model to evaluate for an additional effect. The incremental benefit of T2CL was assessed
by improvement in the area under the receiver-operating curve (AUC) before and after
its addition to the multivariate model. Cohen’s kappa coefficient was used to test
the interobserver variability of CMR wall motion pattern and LV lead position. Statistical
significance was determined by p??0.05 on two-tailed analysis.