J Interv Card Electrophysiol (2012) 35:311–321 DOI 10.1007/s10840-012-9709-y

Sinus rhythm electrocardiogram identification of basal-lateral ischemic versus nonischemic substrate in patients with ventricular tachycardia Brian P. Betensky & Marc W. Deyell & Wendy S. Tzou & Erica S. Zado & Francis E. Marchlinski Received: 19 March 2012 / Accepted: 21 June 2012 / Published online: 12 August 2012 # Springer Science+Business Media, LLC 2012

Abstract Purpose Sinus rhythm (SR) electrocardiogram (ECG) features in patients with nonischemic cardiomyopathy (NICM) and ventricular tachycardia (VT) have been described. ECG characteristics that distinguish nonischemic VT substrate from prior myocardial infarction (MI) have yet to be determined. We aimed to identify ECG differences between patients with basal-inferolateral scar due to NICM versus prior MI. Methods SR/atrial-paced ECGs from patients who underwent VT ablation with endocardial/epicardial basalinferolateral nonischemic scar (n025) were compared to patients with inferior/inferolateral MI (n 030). Surface QRS complexes in each lead were analyzed. Patients with bundle branch block or ventricular pacing were excluded. The best diagnostic algorithm was determined by multivariate analysis then validated prospectively. Results The NICM group had smaller R amplitude in leads I, II, and III (p≤0.05 for all), greater S amplitude in leads II, III, and V6 (p≤0.001 for all) and S/R ratio in lead V6 (p0 0.001). Inferior Q waves were uncommon in NICM (24 % vs. 87 %, p<0.001). Lateral QRS fragmentation was uncommon (20 %) but only found in NICM. A three-step algorithm was derived with 100 % sensitivity and 77 % specificity for NICM. In the validation cohort (n 051), ICM was appropriately excluded in 93 % of the cases of NICM (91 % interobserver agreement) by the algorithm. Conclusions Lateral lead QRS fragmentation, absence of inferior Q waves, and lead V6 S/R ratio ≥0.25 on the SR ECG distinguishes patients with basal–lateral scar due to B. P. Betensky : M. W. Deyell : W. S. Tzou : E. S. Zado : F. E. Marchlinski (*) Hospital of the University of Pennsylvania, 3400 Spruce Street, Founders 9, Philadelphia, PA 19104, USA e-mail: [email protected]

NICM from those with prior MI. These findings demonstrate the value of the surface ECG in identifying unique scar-based VT substrate. Keywords Electrocardiogram . Cardiomyopathy . Tachyarrhythmias . Catheter ablation Abbreviations VT Ventricular tachycardia NICM Nonischemic cardiomyopathy ECG Electrocardiogram SR Sinus rhythm LVEF Left ventricular ejection fraction BMI Body mass index PWTDd Posterior wall thickness in diastole AAD Antiarrhythmic drugs EAM Electroanatomic map

1 Introduction In the setting of remote myocardial infarction (MI), the substrate for scar-based reentrant ventricular tachycardia (VT) is characterized by relatively well demarcated, predominantly subendocardial fibrosis [1–3]. The arrhythmogenic substrate in nonischemic cardiomyopathy (NICM) is usually histopathologically distinct from ischemic scar, consisting of patchier fibrosis and myocyte structural abnormalities [4, 5]. These latter pathologic changes have a predilection for not only endocardial but also variably to the mid-myocardial and epicardial layers of the basilar and perivalvular region for reasons that remain unclear [6–8]. In cases of transmural extension, it can be particularly challenging to distinguish these two types of scar using current diagnostic techniques. The scar burden in NICM characteristically extends from the perivalvular regions of the left

312

ventricle toward the apex with typical involvement of the basal inferolateral aspects of the left ventricle (LV), mimicking an inferior and/or inferoposterior infarction. In addition, although the extent of endocardial bipolar voltage abnormalities tend to be smaller in NICM, endocardial voltage map characteristics may appear like that of prior infarction making characterization of the underlying substrate challenging even in the electrophysiology laboratory. In these select cases, substrate definition becomes critically dependent on assessment of the functional status of the coronaries and/or a history consistent with prior infarction. However, not uncommonly, some non-occlusive coronary disease may coexist making the etiology for the nonischemic substrate incorrectly attributed to a prior ischemic process. It would be helpful to have corroborative evidence. In patients with VT who are potential candidates for catheter ablation, knowledge of the underlying substrate is an essential step in the development of a successful ablation strategy. While there has been increasing use of delayed enhancement cardiac magnetic resonance imaging, it is expensive, poorly tolerated by some patients, and few centers use the study routinely in patients with implantable defibrillators [9, 10]. Furthermore, clearly distinguishing ischemic versus nonischemic scar using MR imaging is not always possible. Alternatively, the surface electrocardiogram (ECG) is routinely performed, inexpensive, well-tolerated, and has been shown to be effective for the assessment and quantification of scars in patients with cardiomyopathy [11, 12]. We have recently described 12-lead ECG features associated with nonischemic scar in the basal–lateral distribution when compared to patients with NICM but no VT [13]. However, ECG criteria have not yet been established for distinguishing ischemic versus nonischemic scar in the same anatomic distribution. Herein, we present a detailed comparison of the SR 12-lead ECG in patients with a history of VT due to nonischemic basal inferolateral scar and those with basal inferolateral scar due to remote myocardial infarction. The primary aim of this study was to identify characteristics of the SR 12-lead ECG unique to nonischemic and ischemic cardiomyopathy in patients with VT. These data may provide adjunctive diagnostic information for patients where the underlying etiology is unclear, improve our understanding of the impact of scar on sinus rhythm activation, and potentially contribute to the development of an ablation strategy when undertaking a catheterbased treatment approach.

2 Methods 2.1 Patient selection All subjects were referred for VT ablation at the Hospital of the University of Pennsylvania (1999–2010). All subjects

J Interv Card Electrophysiol (2012) 35:311–321

included in the current analysis gave written informed consent for catheter ablation in accordance with the institutional guidelines of the University of Pennsylvania Health System. Subjects had VT in the setting of NICM or ischemic cardiomyopathy (ICM) and (1) no bundle branch block in the sinus or atrial-paced rhythm and (2) detailed LV endocardial and/or epicardial voltage maps with basal inferolateral areas (>6 cm2) of LV low-voltage consistent with scar. Bundle branch block was defined by standard published criteria for the interpretation of the ECG [14]. ICM patients were defined by LV systolic dysfunction with regional wall motion abnormalities on echocardiogram or nuclear stress test, coronary angiography demonstrating right coronary artery or left circumflex artery occlusion, and history of inferior/ inferolateral myocardial infarction defined by standard ECG and laboratory testing with or without coronary artery bypass grafting. NICM patients had diminished LV ejection fraction (<50 %) in the absence of clinically significant coronary artery disease or prior myocardial infarction. Degree of LV hypertrophy was extracted from echocardiography reports in the electronic medical record. 2.2 Electroanatomic mapping protocol Patients were brought to the EP laboratory after consent was obtained. Endocardial voltage maps were created using the CARTO® electroanatomical mapping system (Biosense Webster, Diamond Bar, CA, USA). A bipolar signal amplitude of <0.5 mV demarcated areas of dense scar; those >1.5 mV corresponded with normal tissue; and those 0.5–1.5 mV were used to determine border zone areas between scar and normal tissue [15]. Low endocardial voltage areas indicative of scar were measured offline using the standard surface area measurement tool on the CARTO® 3 system v2.3 and CARTO® v9.6.34. In patients where endocardial ablation was unsuccessful and/or epicardial substrate was highly suspected, epicardial mapping and ablation were pursued as per the operator’s discretion. In the latter cases, access to the pericardium was obtained using the well described Sosa technique [16]. LV epicardial scar surface area was determined by the region of contiguous abnormal electrograms, defined by bipolar electrogram peak-to-peak voltage <1.0 mV associated with prolonged (>80 ms), split (≥2 components separated by ≥20 ms isoelectric interval), or isolated late (occurring after the surface QRS complex) potentials [17, 18]. After interpreting recorded electrograms in the region of low voltage, the area was encircled on the 3-dimensional virtual geometry using the incorporated CARTO® area measurement tool yielding the calculated surface area. Tagged valvular structures and sites directly under epicardial vessels were excluded from analysis. The course of the epicardial coronary vessels were delineated based on coronary

J Interv Card Electrophysiol (2012) 35:311–321

313

To develop clinically useful criteria that could be applied outside the EP laboratory, measurements were made on the sinus- or atrial-paced QRS complexes recorded on routine 12-lead ECGs (frequency range, 0.5–150 Hz, 25 mm/s, 10 mm/mV) obtained from all patients. For quantitative measures, leads I, II, and III in the frontal plane and leads V1 and V6 in the transverse plane were selected for analysis. Inferior Q waves were assessed in leads II, III, and aVF. Lateral Q waves were assessed in leads I, avL, V5, and V6. QRS notching/fragmentation was assessed in all leads except aVR. To account for subtle electrocardiographic differences caused by slight variations in lead placement and fluctuations in QRS amplitude due to the respiratory cycle, two different pre-procedural ECGs recorded on different days were selected, and the average of the two measurements was then used for analysis. Standard 12-lead ECG electrode positions were identified and lead placement

performed by skilled technicians. R- and S-wave amplitudes were measured utilizing the PR and/or ST segments, respectively, as the isoelectric reference baseline (Fig. 1). R/S and S/R ratios then were calculated. Both limb and precordial lead voltages were measured as peak-to-peak (R + S) amplitudes. Low limb–lead voltage was defined as a peak-to-peak (R + S) wave amplitude of <0.5 mV in all of the limb leads, and low precordial–lead voltage was defined using a similar measurement of <1.0 mV in the precordial leads [19]. Pathologic Q waves were defined as an initial negative deflection to the QRS complex ≥0.03 s and >0.1 mV deep or a QS in two contiguous limb leads. Total QRS duration was measured in the 12-lead rhythm format from the earliest initial deflection from the isoelectric line in any lead to the time of latest offset in any lead. Individual leads were then similarly measured to assess for regional differences (Fig. 1). R wave duration was measured from onset of QRS to where the descending component of the R wave intersected the isoelectric line while S wave duration was measured from the R intersection to where the QRS returned to baseline (Fig. 1). QRS notching was counted when there was a departure in both slope and sign from the primary ECG curve not crossing baseline, excluding fundamental directional changes in the QRS complex, in any component

Fig. 1 ECG features studied. (a) quantification of QRS amplitudes and durations; a 0 R-wave duration; b 0 S-wave duration; c 0 QRS duration; d 0 R-wave amplitude; e 0 S-wave amplitude; f 0 QRS

amplitude. (b) Measurement of qualitative QRS abnormalities, including simple R wave notching (1), simple S wave notching (2), and variations of surface QRS fragmentation (3–5)

arteriography performed at the time of epicardial mapping and course further defined based on low-voltage recordings in the absence of split, wide, late, or fractionated electrograms in the same anatomic distribution [17]. 2.3 ECG measurement protocol

314

of the QRS complex in at least one regional lead (Fig. 1(b), 1–2). A QRS complex was considered "fragmented" when it displayed more complex notching, including various multiphasic RSR patterns with or without Q waves, ≥2 R´ or the presence of numerous high frequency deflections in at least one regional lead (Fig. 1(b), 3–5) [20–22]. To validate our criteria, two observers blinded to the substrate diagnosis applied the criteria to a second group of patients who presented for VT ablation between 2010 and 2012. 2.4 Statistical analysis Descriptive variables were reported as mean±standard deviation or proportions. Univariate comparisons were made using Student’st test for continuous variables and Fisher’s exact test or a chi-square test for categorical variables. Variables significant at the 0.05 level were selected for sensitivity and specificity analysis. Receiver operating characteristic (ROC) curves were generated and those with area under the curve (AUC)>0.8 were selected for diagnostic modeling. Multivariate logistic regression analysis was performed including these variables in a forward, stepwise manner. Variables with perfect diagnostic accuracy were excluded from the model. Only those variables that were independently predictive in the final model were incorporated into a dichotomized, unified algorithm to determine the highest cumulative sensitivity and specificity for predicting NICM substrate. Continuous nonnormal variables were logarithmically transformed. Model fit and accuracy was determined by examining each ROC curve for the final model. The Hosmer–Lemeshow test was used to test for goodness of fit. In the validation cohort, Cohen’s kappa was used to test dichotomous variables. Spearman’s rank correlation coefficient was used to quantify agreement between two independent raters. Agreement between the continuous ECG measurements of the blinded observers was also assessed with the intraclass correlation coefficient. All statistical analyses were performed using STATA v9.2 (StataCorp, College Station, TX). Subgroup analysis was performed between ICM patients with prior inferior/inferolateral infarction and NICM patients with similarly distributed inferior/inferolateral endocardial scar to test whether our criteria maintained its predictive strength between these two anatomically matched groups.

3 Results 3.1 Retrospective analysis Thirty subjects with ICM due to documented prior inferior/ inferolateral myocardial infarction were compared to 25 subjects with idiopathic dilated NICM. Coronary angiography

J Interv Card Electrophysiol (2012) 35:311–321

demonstrated right coronary artery (RCA) occlusion (65 %), left circumflex artery (LCX) occlusion (25 %), or complete occlusion or >90 % stenosis of both RCA and LCX (10 %). Sixteen (53 %) underwent revascularization and 14/30 (47 %) were not revascularized. Patient demographics were similar at baseline, except NICM subjects were younger (Table 1). 3.1.1 Electroanatomic mapping characteristics All patients with ICM underwent endocardial mapping and none required epicardial mapping to achieve successful VT ablation. Within the NICM group, all patients underwent endocardial mapping, 16/25 (64 %) of whom demonstrated low voltage consistent with scar. Thirteen patients (52 %) with NICM in this study group underwent epicardial mapping. Comparison of the two groups showed that the endocardial scar area was smaller in the NICM group compared to the ICM group (p00.004) (Table 2). Mean LV epicardial scar area was 69.3±38.2 cm2 in the NICM group with no comparison in the ICM group owing to lack of the need for epicardial mapping (Table 2). While some patients had substantial overlap in endocardial scar distribution between the two groups, patients with ICM tended to have more homogenous scar that was often well-defined and identified in the true posterior and inferoseptal regions (Fig. 2). In contrast, those with NICM often exhibited perivalvular abnormalities with irregular patchy areas extending further to Table 1 Baseline characteristics

Age (years) Males (%) BMI (kg/m2) AADs (%) Amiodarone Quinidine Procainamide Mexilitine Sotalol Flecainide Disopyramide Propafenone LVEF (%) Hypertrophy (%) None Mild Moderate Severe PWTDd (cm)

NICM (n025)

ICM (n030)

p

56±17 80

67±10 93

0.005 0.225

28±5

29±5

0.498

52 4 16 16 20 1 1 1 37±13

57 10 3 23 17 0 0 0 34±9

0.790 0.617 0.165 0.736 1.000 0.455 0.455 0.455 0.272

65 25 5 0 0.92±0.22

55 25 20 0 1.03±0.23

0.582 0.761 0.204 1.000 0.182

BMI body mass index, AADs antiarrhythmic drugs, LVEF left ventricular ejection fraction, PWTDd posterior wall thickness in diastole

J Interv Card Electrophysiol (2012) 35:311–321

315

Table 2 Electroanatomic mapping data in the derivation cohort

2

Bipolar electrogram-defined scar area (cm ) % LV scar Total LV surface area (cm2)

ICM

NICM

ENDO 40.4±21.8 18.0±9.6 224.2±48.1

ENDO 16.5±16.4 8.2±7.6 214.8±40.3

p

ICM

NICM

0.004 0.009 0.603

EPI –a –a –a

EPI 69.3±38.2 30.9±23.1 224.0±58.3

ENDO endocardial, EPI epicardial a

Not mapped

the basal–lateral wall, and at times also involving the basal septum (Fig. 3). In the select NICM patients that underwent epicardial mapping, a scar was often prominent in the epicardial region corresponding not only to the endocardial scar, but in some, was more extensive than the endocardial substrate and was associated with large areas of confluent endocardial unipolar voltage abnormality [23].

With respect to duration and QRS morphology measurements, NICM patients had fewer pathologic inferior Q waves (p<0.001). The NICM group exhibited longer QRS duration in lead I (p<0.001), V1 (p00.037), and V6 (p0 0.016). In the lateral leads, QRS fragmentation only occurred in the NICM group (p 00.015), although simple notching was present in both groups. 3.1.3 Diagnostic algorithm

3.1.2 ECG characteristics ECG measurement data are presented in Table 3. Typical SR ECG tracings for each group are illustrated in Fig. 4. In the frontal plane, amplitude measurements demonstrated that patients with NICM had smaller R amplitude in leads I (p 00.021), II (p 00.015), and III (p 00.051), greater Swave amplitude in lead II (p<0.001) and III (p00.001) and smaller R/S ratios in leads II (p00.016) and III (p0 0.004). In the precordial leads, patients with NICM had greater R amplitude in V1 (p00.055) and V6 (p00.041), S waves in lead V6 (p<0.001), R/S ratio in lead V1 (p00.051) and S/R ratio in lead V6 (p00.001). Fig. 2 Anatomic overlap of endocardial VT substrate due to previous MI versus NICM. Electroanatomic maps showing posteroanterior/left lateral view and low-voltage (<1.5 mV) areas of the LV endocardium. (a, b) Two patients with prior inferolateral infarction; (c, d) Two patients with NICM. Note, in NICM patients, the region of endocardial low bipolar voltage can appear similar to that of patients with prior infarction

Multivariable analysis was performed on the parameters that were statistically significant on univariate analysis. Of these, lateral lead QRS fragmentation, lack of inferior Q waves and lead V6 S/R were the only independent predictors of NICM. Lead II S amplitude, although was significant by t test between the two groups and had an AUC >0.8, it was not found to be an independent predictor and thus removed from our model. Since V6 S/R was the best continuous variable (sensitivity 76 %, specificity 87 % alone), we tested different cutoffs in forward stepwise combination with pertinent categorical variables to determine the most robust diagnostic algorithm (Figs. 5 and 6). Incorporating V6 S/R≥0.25 into

316

J Interv Card Electrophysiol (2012) 35:311–321

Fig. 3 Representative voltage maps of patients with NICM and VT. Electroanatomic maps from two patients with NICM showing areas of low endocardial (<1.5 mV) and corresponding low epicardial bipolar voltage (<1.0 mV) in posterior view. (a) Patient has similar areas of patchy endocardial and epicardial low voltage with limited perivalvular distribution. (b) Patient with the large area of epicardial low voltage that exceeds that of the endocardium and extends from base to apex

the algorithm, we excluded basal inferolateral VT substrate due to ICM in 25/25 (100 %) NICM cases. The algorithm falsely identified 7/30 ICM patients, resulting in a cumulative sensitivity of 100 %, specificity 77 %, negative predictive value of 100 %, and positive predictive value of 78 %. 3.1.4 Subgroup analysis: basal–lateral endocardial scar in matching distribution Of the 25 NICM patients, 15 patients demonstrated sizeable basal–lateral endocardial scar in a matching distribution with the inferior/inferolateral infarction group. Endocardial low bipolar voltage-defined absolute scar area (20.8±21.5 vs. 36.9±23.5 cm2, p00.090) and % LV surface area (16.5± 10.4 vs. 18.5 ± 9.0 cm2, p 00.634) were not statistically different. The derived three-step algorithm correctly ruled out ICM VT substrate in all 15/15 NICM patients and ruled in 23/30 ICM patients, confirming a sensitivity of 100 % and specificity of 77 %. 3.2 Prospective analysis Following the completion of our algorithm, 106 subsequent consecutive VT cases with basal–lateral nonischemic or ischemic substrate were evaluated. After excluding cases that did not meet inclusion criteria (10 LBBB , 5 RBBB, 36 ventricular paced, 3 mixed etiology, 1 scar area <6 cm2), 51 cases meeting our inclusion criteria remained (28 NICM, 23 ICM). Using our algorithm, Reviewer 1 correctly classified 76 % all cases and appropriately ruled out ICM in 93 % NICM patients. Reviewer 2 correctly classified 71 % and ruled out ICM in 86 % NICM patients. Thus, the algorithm

had high sensitivity (93 %) and negative predictive value (87 %) but only modest specificity (57 %) and positive predictive value (72 %) in the prospective cohort. The interobserver agreement in measurement of V6 S/R ratios yielded an intraclass correlation coefficient of 0.850 (95 % confidence interval 0.751–0.911, p<0.001). Spearman’s correlation for V6 S/R was significant (rho00.905). For dichotomous variables, interobserver agreement for lateral QRS fragmentation was 84 % (K00.414) and inferior Q waves 90 % (K00.801). The kappa statistic for the overall model was K00.696 with 91 % inter-rater agreement.

4 Discussion We present new 12-lead SR ECG criteria for distinguishing patients with a history of VT and basal–lateral/inferolateral scar due to a nonischemic versus ischemic substrate. ECGs from patients with NICM compared to ICM exhibited quantifiable amplitude, duration, and morphologic QRS differences. Characteristics in favor of nonischemic etiology were the presence of lateral lead QRS fragmentation, lack of pathologic Q waves in the inferior limb leads and/or greater lead V6 S/R amplitude ratios. While there was a trend towards longer QRS durations in leads I, V1, and V6 and greater inferior lead S wave amplitude in the NICM group, these parameters were not independent predictors in multivariate analysis. Such differences in the SR ECG, suggest that independent of the general anatomic distribution of scars, there may be alterations of the SR wavefront that are unique to nonischemic scars. Recognition of SR ECG features in patients without BBB or ventricular paced

J Interv Card Electrophysiol (2012) 35:311–321

317

Table 3 ECG measurement data NICM Amplitude Limb lead voltage (mV) Precordial lead voltage (mV) Lead I R amplitude (mV) S amplitude (mV) R/S ratio Lead II R amplitude (mV) S amplitude (mV) R/S ratio Lead III R amplitude (mV) S amplitude (mV) R/S ratio Lead V1 R amplitude (mV) S amplitude (mV) R/S ratio Lead V6 R amplitude (mV) S amplitude (mV) S/R ratio Duration Total QRS duration (ms) Lead I QRS duration (ms) R wave duration (ms) S wave duration (ms) Lead III QRS duration (ms) R wave duration (ms) S wave duration (ms) Lead V1 QRS duration (ms) R wave duration (ms) S wave duration (ms) Lead V6 QRS duration (ms) R wave duration (ms) S wave duration (ms) Q waves/QRS notching Inferior Q waves (# cases) Lateral Q waves (# cases) Leads I, aVL, V5, V6 notching Leads I, aVL, V5, V6 Fragmentation Leads II, III, aVF notching Leads II, III, aVF fragmentation Leads V1, V2 notching

ICM

p

0.67±0.28 1.26±0.65

0.73±0.37 1.25±0.61

0.276 0.954

0.6±0.3 0.2±0.1 14.7±15.5

0.7±0.3 0.1±0.1 17.0±29.8

0.021 0.260 0.534

0.4±0.2 0.3±0.2 4.8±15.0

0.5±0.2 0.1±0.1 17.1±20.4

0.015 <0.001 0.016

0.2±0.1 0.5±0.4

0.3±0.2 0.2±0.3

0.051 0.001

1.9±4.3

10.8±14.3

0.004

0.3±0.2 0.9±0.4 0.4±0.3

0.2±0.1 0.8±0.3 0.2±0.2

0.055 0.589 0.051

0.7±0.4 0.3±0.1 0.5±0.6

1.0±0.4 0.1±0.1 0.1±0.1

0.041 <0.001 0.001

117.7±17.0

112.2±14.5

0.301

120±30 70±40 50±30

90±20 60±20 30±20

<0.001 0.213 0.002

130±30 40±20

120±20 50±30

0.169 0.084

80±30

60±30

0.016

120±20 40±10 90±30

110±20 30±10 80±20

0.037 0.379 0.158

120±30 60±30 60±20

100±20 60±20 40±30

0.016 0.720 0.003

6 7 9 5 17 5 4

26 9 8 0 23 1 2

<0.001 1.000 0.562 0.015 0.551 0.082 0.394

318

J Interv Card Electrophysiol (2012) 35:311–321

Table 3 (continued)

NICM Leads V1, V2 fragmentation Leads V3, V4 notching Leads V3, V4 fragmentation

rhythms may be clinically useful in patients with nonocclusive coronary disease where a diagnosis remains uncertain, help to characterize the anticipated substrate in patients with VT and, potentially aid the development of an ablative strategy, i.e., epicardial mapping, when undertaking an ablative treatment approach. In a previous investigation, Tzou et al. compared ECG characteristics among NICM patients with VT to those with NICM and no VT. They demonstrated that V1 R amplitude >0.15 mV and V6S amplitude ≥0.15 mV or V6 S/R ratio ≥0.2 were reliable predictors of basal–lateral nonischemic scar in patients with VT compared to those with NICM and no VT, as indexed by bipolar voltage mapping (sensitivity 86 %, specificity 88 %) [13]. In the present study, we found that lead V1 was not as predictive for identifying nonischemic versus ischemic substrate but that lead V6 was still useful. That V1 R amplitude was not an independent predictor of NICM is not surprising, as the development of prominent R waves in V1 has long been recognized as a hallmark of basal posterior infarction, a Q wave equivalent. A deep S wave and a large S/R ratio in lead V6 were effective for identifying patients with NICM and VT, as well as distinguishing nonischemic scar from ischemic scar in the same territory, suggesting that these may represent true substratespecific changes in the magnitude and direction of the

Fig. 4 ECG features in nonischemic versus ischemic cardiomyopathy. Left: ECGs from five patients with NICM exhibiting variations of lateral lead QRS fragmentation, lack of inferior Q waves, and V6 S/R ratios >0.25. (1) multiphasic qRsr's'r'' morphology; (2) rSr' with terminal QRS fragmentation; (3) fragmented QRS upstroke; (4) broad, fragmented upstroke, and notched downstroke; (5) notched R wave with multiple S wave notches appearing as terminal QRS fragmentation in the lateral leads. Right: ECGs from five patients with prior infarction exhibiting inferior Q waves, small V6 S/R ratios and lack of lateral lead QRS fragmentation

0 5 1

ICM 0 4 0

p 1.000 0.717 1.000

electrical vector [11]. While the precise reasons for this are not known, we speculate this may be due to loss of basal– lateral, late in the QRS, activation forces owing to more extensive epicardial versus endocardial involvement of the lateral wall in NICM compared to ICM group. The absence of pathologic inferior Q waves demonstrated modest sensitivity (76 %) and specificity (87 %) for NICM when assessed as an independent predictor. The high prevalence of Q waves in the inferior frontal plane in ICM suggests greater transmural scar extension in this population, as opposed to NICM in which the endocardium is often spared. We suspect that transmural infarct extension was common among other patients in the ICM group due to the lack of coronary revascularization in 47 %. In our study sample, only 64 % of the NICM group demonstrated basal– inferolateral endocardial voltage abnormalities whereas all patients in the ICM group had dense endocardial lowvoltage areas, nearly double the area of the NICM group when expressed as a percentage of LV endocardial surface area (Table 2). Epicardial voltage abnormalities were a consistent finding in the NICM patients who underwent epicardial mapping and ablation (52 % of the retrospective arm and 86 % of the prospective arm) (Fig. 3). Thus, our findings would suggest that sizable, dense transmural contiguous scar is not required for the VT substrate in NICM.

J Interv Card Electrophysiol (2012) 35:311–321

319

Fig. 5 Receiver–operator characteristic curve and scatter plot illustrating lead V6 S/R amplitude ratios. Left: Sensitivities and specificities for predicting basal–lateral VT substrate using lead V6 S/R amplitude ratio alone are indicated for different cutoffs in the derivation cohort (AUC0 0.919, p<0.001). Note, a ratio of ≥0.25 was the most effective cutoff for the three-step algorithm. Right: Corresponding scatter plot for calculated V6 S/ R ratios in the derivation cohort (black diamonds) and the prospective cohort (asterisks)

With regard to differences in surface ECG timing and intraventricular conduction delay, we identified leadspecific prolongation of the QRS complex in NICM relative to those with ICM in the absence of bundle branch block. In leads I, III, and V6, we noted a significant increase in S wave duration but not in R wave duration when comparing NICM to ICM patients. One plausible explanation for these observations could be that the SR activation wavefront is delayed as it spreads through the abnormal basal–lateral endocardium, typically the latest area to be activated. If this were the result of a relatively more diffuse process of interstitial fibrosis, beyond the basal–lateral region, we might expect loss of limb and precordial voltage in the NICM group relative to the ICM group, which was not a consistent finding when individual leads were averaged (Table 3). Alternatively, since the timing of epicardial activation occurs later in the QRS complex than endocardium, it is also plausible that a prolonged S wave is evidence for delayed transmural activation from the endocardium to

epicardium due to a disease process predominating in the intramural layers of the ventricular myocardium. Identification of notching and QRS fragmentation was pursued as a potential marker of differences in delayed and asynchronous conduction. These abnormalities have been observed in patients with ARVD, Brugada syndrome, coronary artery disease, and cardiac sarcoidosis [22, 24–28]. Surface QRS fragmentation has been suggested as an epsilon wave equivalent in patients with idiopathic dilated NICM, perhaps related to inhomogeneous and disrupted activation of diseased myocardium [29]. Consistent with prior studies, our results revealed QRS notching to be most common in the inferior leads of both patients with ICM and NICM with VT related to basal–lateral substrate and was not a distinguishing feature [21]. Interestingly, surface QRS fragmentation was rare overall and in the lateral leads only occurred in NICM, perhaps owing to more lateral extension of electrogram abnormalities, ventricular dyssynchrony and/ or delayed transmural conduction [30]. Thus, lateral lead surface QRS fragmentation had 100 % specificity for NICM substrate in this study. When a composite of notching plus fragmentation was analyzed, this finding was also more common in the lateral leads of patients with NICM with a nonsignificant trend towards significance (56 % vs. 27 %, p00.052). 4.1 Study limitations

Fig. 6 Three-step diagnostic algorithm. Stepwise assessment of lateral lead QRS fragmentation, lack of inferior Q waves and V6 S/R ratio for discriminating between ICM and NICM VT substrates. Stepwise sensitivities and specificities from the derivation cohort are noted

We performed a retrospective analysis in a relatively welldefined cohort of patients undergoing VT ablation for refractory ventricular arrhythmias, which may have limited our power to detect additional subtle electrocardiographic differences. While not all patients included in this study had precisely the same size or transmural distribution of low-voltage regions consistent with scar, this diversity may make our

320

results more generalizable and applicable to the clinical imperative of distinguishing ischemic versus nonischemic substrate in patients referred for a VT ablation. We also performed a subgroup analysis to confirm the merit of our defined ECG characteristics for patients with directly overlapping basal– lateral endocardial scar. Additionally, because we studied patients with inferior/inferolateral MIs, our criteria may not apply to patients with concomitant infarctions in other anatomic distributions. Lastly, since precordial amplitude is subject to variation in body habitus, electrode positioning, respiration, and cardiac rotation, our findings may be less applicable in patients who demonstrate significant extremes with regard to these parameters.

5 Conclusion Patients with VT and basal–lateral/inferolateral scar possess SR ECG characteristics that reflect the unique scar-based VT substrate. The SR 12-lead ECG in patients with NICM is more likely to manifest lateral lead QRS fragmentation, lack inferior Q waves, and have more marked V6 S/R ratios than the ECG in patients with ICM. Recognition of these features on the routine 12-lead ECG may aid in accurately identifying the VT anatomic substrate.

Funding sources Supported in part by the F. Harlan Batrus Research Fund and the Murray and Susan Bloom Research Fund.

Disclosures None

Conflicts of interest None

References 1. Fenoglio, J. J., Jr., Pham, T. D., Harken, A. H., Horowitz, L. N., Josephson, M. E., & Wit, A. L. (1983). Recurrent sustained ventricular tachycardia: structure and ultrastructure of subendocardial regions in which tachycardia originates. Circulation, 68(3), 518– 533. 2. Kottkamp, H., Wetzel, U., Schirdewahn, P., Dorszewski, A., Gerds-Li, J. H., Carbucicchio, C., et al. (2003). Catheter ablation of ventricular tachycardia in remote myocardial infarction: substrate description guiding placement of individual linear lesions targeting noninducibility. Journal of Cardiovascular Electrophysiology, 14(7), 675–681. 3. Stevenson, W. G., Stevenson, L. W., Middlekauff, H. R., & Saxon, L. A. (1993). Sudden death prevention in patients with advanced ventricular dysfunction. Circulation, 88(6), 2953–2961. 4. Roberts, W. C., Siegel, R. J., & McManus, B. M. (1987). Idiopathic dilated cardiomyopathy: analysis of 152 necropsy patients. The American Journal of Cardiology, 60(16), 1340–1355.

J Interv Card Electrophysiol (2012) 35:311–321 5. Unverferth, D. V., Baker, P. B., Swift, S. E., Chaffee, R., Fetters, J. K., Uretsky, B. F., et al. (1986). Extent of myocardial fibrosis and cellular hypertrophy in dilated cardiomyopathy. The American Journal of Cardiology, 57(10), 816–820. 6. Hsia, H. H., Callans, D. J., & Marchlinski, F. E. (2003). Characterization of endocardial electrophysiological substrate in patients with nonischemic cardiomyopathy and monomorphic ventricular tachycardia. Circulation, 108(6), 704–710. 7. Marchlinski, F. E. (2007). Perivalvular fibrosis and monomorphic ventricular tachycardia: toward a unifying hypothesis in nonischemic cardiomyopathy. Circulation, 116(18), 1998–2001. 8. Eckart, R. E., Hruczkowski, T. W., Tedrow, U. B., Koplan, B. A., Epstein, L. M., & Stevenson, W. G. (2007). Sustained ventricular tachycardia associated with corrective valve surgery. Circulation, 116(18), 2005–2011. 9. Bogun, F. M., Desjardins, B., Good, E., Gupta, S., Crawford, T., Oral, H., et al. (2009). Delayed-enhanced magnetic resonance imaging in nonischemic cardiomyopathy: utility for identifying the ventricular arrhythmia substrate. Journal of the American College of Cardiology, 53(13), 1138–1145. 10. Assomull, R. G., Prasad, S. K., Lyne, J., Smith, G., Burman, E. D., Khan, M., et al. (2006). Cardiovascular magnetic resonance, fibrosis, and prognosis in dilated cardiomyopathy. Journal of the American College of Cardiology, 48(10), 1977–1985. 11. Strauss, D. G., & Selvester, R. H. (2009). The QRS complex—a biomarker that "images" the heart: QRS scores to quantify myocardial scar in the presence of normal and abnormal ventricular conduction. Journal of Electrocardiology, 42(1), 85–96. 12. Strauss, D. G., Selvester, R. H., Lima, J. A., Arheden, H., Miller, J. M., Gerstenblith, G., et al. (2008). ECG quantification of myocardial scar in cardiomyopathy patients with or without conduction defects: correlation with cardiac magnetic resonance and arrhythmogenesis. Circulation. Arrhythmia and Electrophysiology, 1(5), 327–336. 13. Tzou, W. S., Zado, E. S., Lin, D., Callans, D. J., Dixit, S., Cooper, J. M., et al. (2011). Sinus rhythm ECG criteria associated with basal-lateral ventricular tachycardia substrate in patients with nonischemic cardiomyopathy. Journal of Cardiovascular Electrophysiology, 22(12), 1351–1358. 14. Surawicz, B., Childers, R., Deal, B. J., Gettes, L. S., Bailey, J. J., Gorgels, A., et al. (2009). AHA/ACCF/HRS recommendations for the standardization and interpretation of the electrocardiogram: part III: intraventricular conduction disturbances: a scientific statement from the American Heart Association Electrocardiography and Arrhythmias Committee, Council on Clinical Cardiology; the American College of Cardiology Foundation; and the Heart Rhythm Society. Endorsed by the International Society for Computerized Electrocardiology. Journal of the American College of Cardiology, 53(11), 976–981. 15. Marchlinski, F. E., Callans, D. J., Gottlieb, C. D., & Zado, E. (2000). Linear ablation lesions for control of unmappable ventricular tachycardia in patients with ischemic and nonischemic cardiomyopathy. Circulation, 101(11), 1288–1296. 16. Sosa, E., Scanavacca, M., d’Avila, A., & Pilleggi, F. (1996). A new technique to perform epicardial mapping in the electrophysiology laboratory. Journal of Cardiovascular Electrophysiology, 7(6), 531–536. 17. Cano, O., Hutchinson, M., Lin, D., Garcia, F., Zado, E., Bala, R., et al. (2009). Electroanatomic substrate and ablation outcome for suspected epicardial ventricular tachycardia in left ventricular nonischemic cardiomyopathy. Journal of the American College of Cardiology, 54(9), 799–808. 18. Tung, R., Nakahara, S., Ramirez, R., Lai, C., Fishbein, M. C., & Shivkumar, K. (2010). Distinguishing epicardial fat from scar: analysis of electrograms using high-density electroanatomic mapping in a novel porcine infarct model. Heart Rhythm, 7(3), 389– 395.

J Interv Card Electrophysiol (2012) 35:311–321 19. Macfarlane, P. W., & Lawrie, T. D. V. (Eds.). (1989). Comprehensive electrocardiology, theory and practice in health and disease (p. 291). New York: Pergamon Press. 20. Das, M. K., Maskoun, W., Shen, C., Michael, M. A., Suradi, H., Desai, M., et al. (2010). Fragmented QRS on twelve-lead electrocardiogram predicts arrhythmic events in patients with ischemic and nonischemic cardiomyopathy. Heart Rhythm, 7(1), 74–80. 21. Das, M. K., & Zipes, D. P. (2009). Fragmented QRS: a predictor of mortality and sudden cardiac death. Heart Rhythm, 6(3 Suppl), S8– 14. 22. Das, M. K., Suradi, H., Maskoun, W., Michael, M. A., Shen, C., Peng, J., et al. (2008). Fragmented wide QRS on a 12-lead ECG: a sign of myocardial scar and poor prognosis. Circulation. Arrhythmia and Electrophysiology, 1(4), 258–268. 23. Hutchinson, M. D., Gerstenfeld, E. P., Desjardins, B., Bala, R., Riley, M. P., Garcia, F. C., et al. (2011). Endocardial unipolar voltage mapping to detect epicardial ventricular tachycardia substrate in patients with nonischemic left ventricular cardiomyopathy. Circulation. Arrhythmia and Electrophysiology, 4(1), 49–55. 24. Das, M. K., Khan, B., Jacob, S., Kumar, A., & Mahenthiran, J. (2006). Significance of a fragmented QRS complex versus a Q wave in patients with coronary artery disease. Circulation, 113 (21), 2495–2501. 25. Peters, S., Trummel, M., & Koehler, B. (2008). QRS fragmentation in standard ECG as a diagnostic marker of arrhythmogenic right ventricular dysplasia-cardiomyopathy. Heart Rhythm, 5(10), 1417–1421. 26. Morita, H., Kusano, K. F., Miura, D., Nagase, S., Nakamura, K., Morita, S. T., et al. (2008). Fragmented QRS as a marker of conduction abnormality and a predictor of prognosis of Brugada syndrome. Circulation, 118(17), 1697–1704. 27. Hatala, R., Savard, P., Tremblay, G., Page, P., Cardinal, R., Molin, F., et al. (1995). Three distinct patterns of ventricular activation in infarcted human hearts. An intraoperative cardiac mapping study during sinus rhythm. Circulation, 91(5), 1480–1494.

321 28. Schuller, J. L., Olson, M. D., Zipse, M. M., Schneider, P. M., Aleong, R. G., Wienberger, H. D., et al. (2011). Electrocardiographic characteristics in patients with pulmonary sarcoidosis indicating cardiac involvement. Journal of Cardiovascular Electrophysiology, 22(11), 1243–1248. 29. Flowers, N. C., Horan, L. G., Tolleson, W. J., & Thomas, J. R. (1969). Localization of the site of myocardial scarring in man by high-frequency components. Circulation, 40(6), 927–934. 30. Basaran, Y., Tigen, K., Karaahmet, T., Isiklar, I., Cevik, C., Gurel, E., et al. (2011). Fragmented QRS complexes are associated with cardiac fibrosis and significant intraventricular systolic dyssynchrony in nonischemic dilated cardiomyopathy patients with a narrow QRS interval. Echocardiography, 28(1), 62–68.

Editorial Commentary This single-center observational study describes the ECG characteristics to identify ischemic versus nonischemic substrate in patients with VT.Retrospective analysis of sinus rhythm/atrial-paced ECGs from patients with endocardial/epicardial basal-inferolateral scar in nonischemic cardiomyopathy (NICM, n025) were compared to patients with prior inferior/inferolateral myocardial infarction (MI, n030).Subtle but consistent quantifiable morphologic QRS differences were identified.A diagnostic algorithm was developed by multivariate analysis and then validated prospectively in a small cohort (n051).Lateral lead QRS fragmentation, absence of inferior Q waves and S/R ratio ≥0.25 in lead V6 distinguishes patients with basal-lateral scar due to NICM from those with prior MI.Identification of such ECG features in patients may be useful for the characterization of VT substrate, pre-procedural planning and potentially aid in the development of ablative strategies.