Lipomatous Metaplasia Facilitates Ventricular Tachycardia in Patients With Nonischemic Cardiomyopathy

Abstract

BACKGROUND

Ventricular tachycardia (VT) substrate in patients with nonischemic cardiomyopathy (NICM) is complex in distribution and intramural location.

OBJECTIVES

This study sought to test the hypothesis that myocardial lipomatous metaplasia (LM) is a vital anatomic substrate for VT corridors in patients with NICM and VT, and that LM stabilizes current propagation in VT corridors.

METHODS

Among 49 patients with NICM in the 2-center INFINITY (Prospective Intra-Myocardial Fat Deposition and Ventricular Tachycardia in Cardiomyopathy) Study, potential VT viable corridors within the myocardial scar and/or LM were computed from late gadolinium enhancement cardiac magnetic resonance images and were registered with electroanatomical maps. Corridors passing through VT entrance, isthmus, and/or exit sites, estimated by entrainment or pace mapping, were defined as VT corridors. LM was separately distinguished from scar using computed tomography. The SD of current amplitude along each corridor was measured.

RESULTS

Compared with 151 non-VT corridors, 35 VT corridors traversed a substantially higher volume of LM, with a median 236.6 mg (IQR: 13.5–903.4 mg) vs 5.8 mg (IQR: 0.0–57.9 mg) (P < 0.001). Among corridors with computable current amplitude, 28 VT corridors exhibited substantially lower current variation along the corridors, with SD 8.0 μA (25th–75th percentile: 6.1–10.3 μA) vs 14.9 μA (25th–75th percentile: 8.5–23.7 μA) among 71 non-VT corridors (P < 0.001). Individual VT circuit sites (95 out 118) were highly colocalized with LM.

CONCLUSIONS

VT circuitry corridors in NICM are more likely to traverse LM and exhibit reduced current amplitude variation compared with bystander corridors.

Graphical Abstract

CENTRAL ILLUSTRATION Current Stability in Corridors Traversing LM

(A1–4) Current amplitude along LGE-derived viable corridors on left ventricular shell. corridors (white band with purple centerline), (B1–4) traversing LM (green area) (corridors 1 and 3, critical to VT) or without traversing LM (corridors 2 and 4, not critical to VT). (C1–4) Corridors magnified. (D1–4) Scatter plots of current amplitude along the corridors. Red arrow = corridor branches; white points = sampled points; red points = ablation sites; small yellow points = pacing sites; special points in corridor 1 = blue points (point #12) = exits of the VT re-entry corridor; black point (point #15) = delayed potential; pink points (not labelled) = isthmuses of the VT re-entry corridor. Special points in corridor3: blue points (points#6,14, and 18) = exits of the VT re-entry corridor; black point (point #20) = delayed potential; small pink points (points #12 and 16) = fractionated potential; small light blue points (point#19) = double potential. The red arrow indicates the location ofthe branch. LGE = late gadolinium enhancement; LM = lipomatous metaplasia; VT = ventricular tachycardia.

Xu L, et al. JACC Clin Electrophysiol. 2024;10(11):2325–2336.

Patients with nonischemic cardiomyopathy (NICM) and left ventricular dysfunction are predisposed to ventricular tachycardia (VT) with relatively high postablation recurrence.¹ Viable myocardial corridors traversing late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) images have been reported to be associated with appropriate implantable cardioverter-defibrillator (ICD) therapies.² Although previous chronic ischemic cardiomyopathy CMR studies have considered LGE to represent a uniformly fibrotic substrate,³ it often represents a heterogeneous mixture of fibrosis and lipomatous metaplasia (LM).⁴ LM, fibrosis, and surviving myocyte corridors through the infarcted area have been histologically observed.⁵ Using computed tomography (CT) scans to distinguish LM from fibrosis within regions enhancing on LGE, we have previously observed that patients with VT are more likely to exhibit LM within the infarcted myocardium.⁴ Unlike ischemic cardiomyopathy, LM in patients with NICM has been studied rarely. The prevalence of LM in patients with NICM was first reported in the explanted heart study by Baroldi et al.⁶ The presence of LM has been reported to be associated with adverse cardiac remodeling in patients with dilated cardiomyopathy,⁷ and adverse outcomes in patients with prior myocarditis.⁸ LM mass has been reported to be associated with ventricular arrhythmias in an NICM cohort eligible for primary prevention with an ICD.⁹ Recently, we have observed that LM mass is independently associated with postablation VT recurrence.¹⁰ The association of LM proximity with VT circuitry in patients with NICM, however, has never been studied. We sought to investigate the association of LM proximity with electrogram characteristics and viable corridor participation in VT circuitry among patients with NICM.

METHODS

Detailed methodology regarding study population, cardiac CT, and CMR acquisition protocols, cardiac CT and CMR imaging analysis, and statistical analysis are described in the Supplemental Methods.

ELECTROPHYSIOLOGICAL STUDY AND ABLATION.

Antiarrhythmic medications were discontinued for 5 half-lives before the procedure (aside from amiodarone, which was discontinued 2 weeks prior apart from urgent ablations, among whom the drug was ineffective at VT suppression). ICD tachyarrhythmia detection was disabled before ablation and restored to the original mode after the procedure. General anesthesia or conscious sedation was used according to clinical requirements. Endocardial and/or epicardial mapping of the ventricles was performed during sinus rhythm or ventricular pacing using a 3-dimensional electroanatomical mapping (EAM) system. Substrate mapping and activation mapping were performed for sustained hemodynamically tolerated VT. For hemodynamically nontolerated VT, pace mapping was employed. Most cases (47 [96%]) were performed using the CARTO system (Biosense Webster, Inc; or EnSite Precision). To ensure proper catheter contact, accurate electrogram annotation and impedance measurement, and accuracy of electroanatomical to image registration, a catheter with 3.5-mm tip electrode, 2-mm proximal electrode, and 2-mm interelectrode spacing was used (Thermocool Smart Touch, Biosense Webster) and points with <5 g contact force were rejected during mapping. During epicardial mapping, Decanav or Pentaray multipolar catheters (Biosense Webster) were used per operator preference. Two or more consecutive electrograms displaying the same morphology were required at each EAM point. Bipolar electrograms were recorded between adjacent tip and proximal ring electrodes, and unipolar electrograms were recorded between the electrode tip and the Wilson Central Terminal. When using the CARTO system, the filtering frequency was set at 16 to 500 Hz and 2 to 240 Hz for bipolar and unipolar electrograms, respectively. For 2 (4%) cases performed using the EnSite Precision EAM system (Abbott), the Advisor HD Grid catheter was used. With the Ensite system, the filtering frequency was set at 30 to 300 Hz and 0.5 to 100.0 Hz for bipolar and unipolar electrograms, respectively. Identification of VT corridors sites required: 1) concealed fusion upon overdrive pacing during VT, stimulus to QRS during pacing within 20 ms of the electrogram to QRS during VT, and a postpacing interval within 20 ms of the tachycardia cycle length; or 2) 12/12 ECG morphology pace-map match to VT and inability to reinduce VT after ablation at the site.¹¹ When possible, such sites were characterized further as entrance (ie, stimulus-QRS interval >70% of VT cycle length), isthmus (ie, stimulus-QRS interval 30%–70% of VT cycle length), or exit (stimulus-QRS interval <30% of VT cycle length) locations. Circuit target sites were ablated using irrigated radiofrequency (Thermocool, Biosense Webster; or Tacticath, Abbott Technologies) energy starting at 30 W titrated up to 50 W for ≥30 seconds but up to 4 minutes while monitoring changes in impedance.

IMPEDANCE MEASUREMENT.

High-resolution impedance measurement circuits were only available in the CARTO system, mainly during endocardial mapping. As previously reported,⁴ simultaneous with endocardial mapping, the built-in circuits measure the impedance during sinus rhythm or ventricular pacing through a unipolar system, including the generator, wires, catheter, skin-patch interface, interposed tissues, and the electrode-tissue interface (2 skin patches placed at both sides of the flank.). The impedance at each site was measured by an alternating current (6.5 μA, 45 kHz, sinusoidal) injected from the catheter tip.

CURRENT MEASUREMENT.

As we previously reported,⁴ Ohm’s law ($I=V/R$) was used to estimate current amplitude at EAM points.^12–14 The impedance and unipolar voltage measurements involved an identical cathode (catheter tip electrode) and similar anodes which minimize resistance (skin patches on both flanks for impedance measurements and Wilson Central Terminal for voltage amplitude measurement). Thus, based on Ohm’s law, the instantaneous current passing the myocardium is approximated closely by the instantaneous unipolar voltage at each site divided by tissue impedance at each site, which is reasonably constant at the same site. As we previously reported,⁴ current fluctuation was defined as the SD of the current at successive points along each corridor due to the potentially bidirectional activation of corridors during sinus rhythm. The fluctuation of the bipolar voltage amplitude and unipolar voltage amplitude was defined in the same manner.

IMAGE REGISTRATION.

The LGE-MRI and CT angiography images were registered to each other and the EAM mesh using rigid registration protocols in ADAS VT software (ADAS 3D Medical SL). One tool uses corresponding anatomical landmarks that the user identifies on both LGE-MRI/CT angiography and EAM maps. Another tool uses a dedicated user interface to manually adjust the position and orientation of the shells with respect to each other to ensure a seamless fit with the shells. The endocardial EAM data was registered with first to the fifth layers of myocardium (50% mid-myocardial depth), and the surface registration error was measured as the distance between the mapped EAM surface and subendocardial surface of the left ventricular mesh. The epicardial EAM data was registered with the sixth to ninth layers (60%–90% depth), and the surface registration error was measured as the distance between the mapped EAM surface, and subepicardial surface of the left ventricular mesh. After coregistration, image and electrogram characteristics at each EAM point were tabulated as panel data clustered by patient and layers of left ventricular shell derived from LGE-CMR and CT images.

DEFINITION OF VT CORRIDORS.

After the LGE-CMR images were registered with the EAM map, the LGE-derived corridors that corresponded to VT sites on EAM were defined as VT corridors. If the VT sites were >3 mm away from the boundary of corridors, these corridors were classified as non-VT corridors.

STATISTICAL ANALYSIS.

Absolute values and z-scores were reported for electrogram characteristics, where $z=(x-\text{μ})/\sigma$, with $\text{μ}$ and $\text{σ}$ as the mean and the SD of electrogram measurements and $x$ as each individual electrogram measurement. Multilevel random effects linear regression models, clustered by patient and by layers of the left ventricular shell, were used to evaluate the association of tissue types with electrogram measurements. Other parts of statistical analysis were detailed in the supplemental method.

RESULTS

BASELINE CHARACTERISTICS, PROCEDURE CHARACTERISTICS, AND INTRACARDIAC ELECTROGRAM INCLUSION.

The baseline characteristics are summarized in Table 1. The cohort comprised 49 patients with a mean age of 60 years and a mean left ventricular ejection fraction of 35%. The procedure characteristics are summarized in Figure 1A. Of 49 patients, 34 (69.4%) required endocardial ablation only, 11 (22.4%) required endocardial and epicardial ablation, and 4 (8.2%) had epicardial ablation only (Figure 1). VT corridor sites, including the entrances, isthmuses, and exits, were predominantly (ie, approximately two-thirds) distributed in the base of the left ventricle (Supplemental Table 1).

TABLE 1. Baseline Characteristics (N = 49: Endocardial Ablation, n = 45; Epicardial Ablation, n = 15).

Vital statistics
Age, y	61 ± 11
Male	40 (81.6)
Body mass index, kg/m2	31 ± 7
Systolic blood pressure, mm Hga	116 ± 18
Diastolic blood pressure, mm Hga	75 ± 11
Heart rate, beats/mina	69 ± 10
Disease history
Heart failure	32 (65.3)
Hypertension	28 (57.1)
Diabetes mellitus	12 (24.5)
Other arrhythmia (AF/AFL or PVC)	29 (59.2)
Pulmonary disease	5 (10.2)
Renal Insufficiency	6 (12.2)
Hyperlipidemia	13 (26.5)
Hypothyroidism	9 (18.4)
Obstructive sleep apnea	8 (16.3)
Presence of ICD before imaging and VT ablation	48 (98.0)
Laboratory tests
Serum creatinine, mg/dLa	1.1 ± 0.3
Antiarrhythmic medications
Amiodarone	20 (40.8)
Sotalol	9 (18.4)
Mexiletine	1 (2.0)
Other cardiovascular medications
Antiplatelet	17 (34.7)
Anticoagulation	21 (42.9)
Statin	25 (51.0)
ACE inhibitor/ARB use	31 (100.0)
Beta-blocker use	30 (61.2)
Mineralocorticoid receptor antagonist	15 (30.6)
Diuretics	23 (46.9)
Etiologies and classification of NICM
Dilated cardiomyopathy	29 (59.2)
Hypertrophic cardiomyopathy	6 (12.2)
Cardiac sarcoidosis	8 (16.3)
Valvular Heart disease	2 (4.1)
Others	4 (8.2)
Cardiac MRI measurements
LVEF, %	35 ± 12
LVEDVi, mL/m2	117 ± 37
LVESVi, mL/m2	78 ± 35
LV Massi, g/m2	72 ± 22
RVEF, %	39 ± 11
RVEDVi, mL/m2	92 ± 22
RVESVi, mL/m2	56 ± 20
BZ, %	45.5 (26.3, 59.0)
Dense LGE, %	15.3 (9.7, 22.0)
Fibrosis, %b	62.7 (35.5, 74.6)
Fibrosis presence	49 (100)
Cardiac CECT measurements
LM mass, g	2.0 (0.8, 6.6)
LM presence	44 (89.8)

Values are mean ± SD, n (%), or median (25th–75th percentile).

^aMeans at the time immediately before VT ablation.

^bFibrosis percentage was calculated by subtracting LM percentage from BZ and dense LGE percentage.

ACEI = angiotensin converting enzyme inhibitor; AF/AFL = atrial fibrillation/flutter; ARB = angiotensin II receptor blocker; ICD = implantable cardioverter-defibrillator; LM = lipomatous metaplasia; LVEF = left ventricular ejection fraction; LVEDVi = left ventricular end-diastolic volume index; LVESVi = left ventricular end-systolic volume index; LV Massi = left ventricular mass index; PVC = premature ventricular contraction; RVEF = right ventricular ejection fraction; RVEDVi = right ventricular end-diastolic volume index; RVESVi = right ventricular end-systolic volume index.

FIGURE 1. Flow Chart of Ablation Procedure Characteristics, EAM, and Tissue Characterization Data.

EAM = electroanatomical mapping; ICD = implanted cardiac devices; IQR = interquartile range; LGE = late gadolinium enhancement; LM = lipomatous metaplasia; NICM = nonischemic cardiomyopathy; NIPS = noninvasive programmed stimulation; PES = programmed electrical stimulation; VT = ventricular tachycardia.

The inclusion and exclusion of EAM points are summarized in Figure 1B. Among 45 patients with endocardial ablation, 22,910 intracardiac points were included in the final analysis (5,288 in healthy myocardium, 10,405 in the border zone (BZ), 5,896 in scar, and 1,320 points in LM). Impedance data was unavailable for 2 Ensite EAM and 7 Carto EAM cases using multipolar catheters; thus, a total of 11,712 EAM points had impedance recorded. Among 15 patients with epicardial ablation, 28,646 intracardiac points were included in the final analysis (9,687 in healthy myocardium, 10,498 in BZ, 5,742 in scar, and 2,719 points in LM).

TISSUE-SPECIFIC INTRACARDIAC ELECTROGRAM CHARACTERISTICS.

In the endocardial EAM data, a significant stepwise decrease in voltage, impedance, and current from healthy myocardium, BZ, to fibrosis was observed. Compared with fibrosis, electrograms within LM regions exhibited decreased bipolar voltage amplitude (mean coefficient β: −0.77 vs −1.09; P < 0.05), unipolar voltage amplitude (mean coefficient β: −2.00 vs −1.64; P < 0.05), and current amplitude (mean coefficient β: −17.07 vs −14.42; P < 0.001). Impedance within LM was significantly higher than scar (median coefficient β: −9.85 vs −11.08; P < 0.001) but lower than the BZ (Table 2).

TABLE 2. Tissue-specific Electrogram Characteristics on EAM.

Electrogram Characteristics	Endocardial EAM (n = 45 patients; 22,910 EAM points)			Epicardial EAM (n = 15 patients; 28,646 EAM points)
	Tissue Types of Myocardium	Absolute Value β (vs Reference)	Z Score β (vs Reference)	Tissue Types	Absolute Value β (vs Reference)	Z Score β vs Reference
Bipolar voltage (mV)	Normal 5,288 (23.1%)	Reference	Reference	Normal 9,687 (33.8%)	Reference	Reference
	BZ 10,405 (45.4%)	−0.56 (−0.73 to −0.39)a	−0.06 (−0.07 to −0.05)a	BZ 10,498 (36.7%)	−0.06 (−0.16 to 0.03) a,b	−0.02 (−0.03 to −0.01)a
	Scar 5,896 (25.7%)	−0.77 (−0.96 to −0.57)a,b	−0.16 (−0.17 to −0.15)a,b	Scar 5,742 (20.0%)	−0.64 (−0.76 to −0.53)a,b	−0.20 (−0.21 to −0.19)a,b
	LM 1,320 (5.8%)	−1.09 (−1.39 to −0.79)a,b,c	−0.18 (−0.20 to −0.16)a,b,c	LM 2,719 (9.5%)	−0.83 (−0.97 to −0.69)a,b,c	−0.28 (−0.30 to −0.27)a,b,c
Unipolar voltage (mV)	Normal 5,288 (23.1%)	Reference	Reference	Normal 9,687 (33.8%)	Reference	Reference
	BZ 10,405 (45.4%)	−1.02 (−1.21 to −0.84)a	−0.13 (−0.14 to −0.12)a	BZ 10,498 (36.7%)	−0.70 (−0.80 to −0.59)a	−0.14 (−0.16 to −0.13)a
	Scar 5,896 (25.7%)	−1.64 (−1.85 to −1.43)a,b	−0.25 (−0.26 to −0.23)a,b	Scar 5,742 (20.0%)	−2.28 (−2.41 to −2.16)a,b	−0.48 (−0.49 to −0.47)a,b
	LM 1,320 (5.8%)	−2.00 (−2.31 to −1.68)a,b,c	−0.27 (−0.30 to −0.25)a,b,c	LM 2,719 (9.5%)	−2.54 (−2.70 to −2.38)a,b,c	−0.50 (−0.51 to −0.48)a,b,c
Impedance (Ω) n = 11,712	Normal 3,317 (28.2%)	Reference	Reference
	BZ 5,184 (44.0%)	−4.63 (−4.90 to −4.37)a	−0.16 (−0.18 to −0.15)a
	Scar 2,495 (21.2%)	−11.08 (−11.40 to −10.77)a,b	−0.51 (−0.53 to −0.49)a,b
	LM 777 (6.6%)	−9.85 (−10.31 to −9.39)a,b,c	−0.46 (−0.48 to −0.43)a,b,c
Currentd (μA) n = 11,712	Normal 3,317 (28.2%)	Reference	Reference
	BZ 5,184 (44.0%)	−7.41 (−8.21 to −6.62)a	−0.12 (−0.13 to −0.10)a
	Scar 2,495 (21.2%)	−14.42 (−15.83 to −13.03)a,b	−0.34 (−0.37 to −0.31)a,b
	LM 777 (6.6%)	−17.07 (−18.02 to −16.10)a,b,c	−0.42 (−0.44 to −0.40)a,b,c

Values are n (%) unless otherwise indicated. Multilevel random effects linear regression models, clustered by patient and by layers of the left ventricular shell, were used to evaluate the association of tissue types with electrogram measurements. Coefficient β indicates that the corresponding group has an average β-unit increase of intracardiac features (voltage, impedance, and current) or corresponding modified z score compared with the reference group, in multilevel tests clustered by patients. Due to the use of multipolar catheter during epicardial mapping, only approximately 3.5% of intracardiac electrograms had impedance recorded, resulting in no availability of impedance and current.

^aP < 0.05 vs normal myocardium.

^bP < 0.05 vs BZ tissue.

^cP < 0.05 vs scar.

^dCurrent amplitude in LM represents the current of the surviving myocardial bundles within LM, which are beyond the resolution of the high-resolution cardiac computed tomography in the current study.

BZ = border zone; other abbreviations as in Table 1.

Due to the unavailability of impedance from epicardial mapping, only bipolar and unipolar voltage amplitude were compared among 4 types of tissues (Table 2), which exhibited a similar trend as the endocardial EAM data.

LM PROXIMITY TO VT CORRIDORS.

A total of 186 corridors were computed from LGE-CMR images and were registered with EAM points from both endocardial EAM data (129 corridors) and epicardial EAM data (57 corridors). Thirty-five of 186 corridors were VT corridors. Compared with non-VT corridors (Central Illustration, Corridors #2 & 4), VT corridors (Central Illustration, Corridors #1 & 3) traversed a substantially higher volume of LM, with a median 236.6 vs 5.8 mg (P < 0.001) (Table 3 and Figure 2A), and higher LM transmurality (median: 100% vs median: 50% for only corridors traversing; P < 0.001) (Table 3). An optimal threshold value of LM mass within the corridor of >44 mg was computed via multilevel logistic regression to differentiate VT from non-VT corridors with 72% sensitivity and 73% specificity (C-statistic: 0.7603) (Figure 2B).

TABLE 3. LM, Current Amplitude and, Unipolar, and Bipolar Voltage Amplitude Variation Along Noncritical Corridors vs Critical VT Corridors.

	Noncritical Corridors (n = 151)	Critical Corridors (n = 35)	P Value
LM volume within the corridors, (mg)	5.8 (0.0–57.9)	236.6 (13.5–903.4)	<0.001
Corridors with >1 mg LM	79 (52.3%)	31 (88.6%)	0.0001
Transmurality of adjacent LM of all corridors	0% (0%–50%)	100% (50%–100%)	<0.001
Transmurality of adjacent LM of corridors traversing LM	50%% (20%–90%)	100% (70%–100%)	<0.001
Corridor mass, g	1.2 (0.4–3.5)	5.5 (0.8–27.3)	<0.001
SD value of bipolar voltage along each corridor	1.3 (0.7–2.2)	0.8 (0.4–1.5)	0.001
SD value of unipolar voltage along each corridor	2.1 (1.2–3.1)	1.6 (1.0–1.8)	0.002
EAM points per corridor	18 (7–92)	34 (6–91)	0.766
	(n = 71)	(n = 28)
SD value of current along each corridor, μA	14.9 (8.5–23.7)	8.0 (6.1–10.3)	<0.001
EAM points per corridor	10 (6–24)	9 (5–75)	0.983
Branches of each corridor	2 (1–4)	1 (0–2)	0.009

Values are median (IQR) or n(%).

Abbreviations as in Table 1.

FIGURE 2. LM Mass Within Corridors.

(A) Violin plot of LM mass within noncritical corridors vs LM mass within critical VT corridors. (B) Receiver operating characteristic curve using multilevel analysis clustered by patients—an optimal threshold value of LM volume as 44 mg within corridor to differentiate the critical VT corridors from noncritical corridors, with sensitivity of 0.74, specificity of 0.73, with area under the curve of 0.7603. *Significantly different from noncritical corridors. Abbreviations as in Figure 1.

CURRENT FLUCTUATION ALONG VT VS. NON-VT CORRIDORS.

The SD of bipolar voltage, unipolar voltage, and current amplitude of consecutive EAM points along each corridor were used to represent their variation along the corridors (Central Illustration). Substantially lower variation of bipolar voltage and unipolar voltage along the corridors was observed in 35 VT corridors when compared with 151 non-VT corridors, with SDs of 0.8 mV (25th–75th percentile: 0.4–1.5 mV) vs 1.3 mV (25th–75th percentile: 0.7–2.2 mV), with P = 0.001, and SDs of 1.6 mV (25th–75th percentile: 1.0–1.8 mV) vs 2.1 mV (25th–75th percentile: 1.2–3.1 mV), with P = 0.002, respectively (Table 3). Current variation was only compared in the corridors registered with endocardial mapping. Compared with 71 non-VT corridors (Central Illustration, Corridors #2 and 4), 28 VT corridors (Central Illustration, Corridors #1 and 3) exhibited substantially lower current variation along the corridors, with 8.0 μA (25th–75th percentile: 6.1–10.3 μA) vs 14.9 μA (25th–75th percentile: 8.5–23.7 μA), with P < 0.001 (Table 3). Notably, non-VT corridors connected with other corridors more frequently compared with VT corridors, with a median of 2 (IQR: 1–4) vs 1 (IQR: 0–2), with P = 0.009 (Table 3, Central Illustration).

LM PATTERNS AND SPECIFIC VT CIRCUIT SITES.

We identified 118 VT corridor sites, with 76.5% adjacent or within LM on endocardial mapping and 86.0% adjacent or within LM on epicardial mapping. Similarly, the isthmus sites and the exit sites of VT corridors are predominantly located adjacent to or within LM in both endocardial and epicardial mapping (Table 4, Figure 3). Furthermore, approximately two-thirds of VT corridor sites were located in the base and approximately one-third in the mid-ventricle; the VT site distribution overlapped LM distribution (Figure 3).

TABLE 4. VT Sites and LM.

Proximity of VT Sites to LM
		Endocardial Ablation			Epicardial Ablation
		Distant from LM		Adjacent or within LM	Distant from LM	Adjacent or within LM
All entrance, isthmus, (endocardial ablation: n = 68; epicardial ablation: n = 50)		16 (23.5)		52 (76.5)	7 (14.0)	43 (86.0)a
Entrance, (endocardial ablation: n = 2; epicardial ablation: n = 4)		2 (100.0)		0 (0.0)b	0 (0.0)	4 (100.0)b
Isthmus, (endocardial ablation: n = 27; epicardial ablation: n = 7)		6 (22.2)		21 (77.8)	0 (0.0)	7 (100.0)a
Exit, (endocardial ablation: n = 39; epicardial ablation: n = 39)		8 (20.5)		31 (79.5)	7 (18.0)	32 (82.1)a
Characteristics of LM Adjacent to or Surrounding Critical VT Sites
	All Entrance, Isthmus, Exit, (Endocardial Ablation: n = 68; Epicardial Ablation: n = 50)			Entrance (Endocardial Ablation: n = 2; Epicardial Ablation: n = 4)
	Subendocardial LM (1st-3rd layers)	Midwall Myocardial LM (4th–6th Layers)	Subepicardial LM (7th-9th Layers)	Subendocardial LM (1st-3rd Layers)	Midwall Myocardial LM (4th–6th Layers)	Subepicardial LM (7th-9th Layers)
Endocardial ablation	29 (42.6)	31 (45.6)	34 (50.0)	0 (0)	0 (0)	0 (0)
Epicardial ablation	23 (46.0)	25 (50)	37 (74.0)c,d	2 (50.0)	2 (50.0)	4 (100)
	Isthmus (Endocardial Ablation: n = 27; Epicardial Ablation: n = 7)			Exit (Endocardial Ablation: n = 39; Epicardial Ablation: n = 39)
	Subendocardial LM (1st-3rd Layers)	Midwall Myocardial LM (4th–6th Layers)	Subepicardial LM (7th-9th Layers)	Subendocardial LM (1st-3rd Layers)	Midwall Myocardial LM (4th–6th Layers)	Subepicardial LM (7th-9th Layers)
Endocardial ablation	12 (44.4)	13 (48.1)	16 (59.5)	17 (43.6)	18 (46.2)	18 (46.2)
Epicardial ablation	4 (57.1)	6 (85.7)	7 (100)	17 (43.6)	17 (43.6)	26 (66.7)c,d

Values are n (%).

^aStatistically significantly different from frequency of critical VT sites distant from LM, with P value < 0.001.

^bStatistically significantly different from frequency of critical VT sites distant from LM, with P value < 0.05.

^cStatistically significantly different from sub-endocardial LM.

^dStatistically significantly different from mid-wall LM.

Abbreviations as in Table 1.

FIGURE 3. Distribution of VT Sites and LM in NICM.

Distribution of (A) VT corridors, (B) VT corridors adjacent to LM, and (C) LM, according to the 17-segment AHA model. AHA — American Heart Association; other abbreviations as in Figure 1.

Among 68 VT sites from endocardial mapping, the adjacent LM was distributed more evenly in the subendocardium, mid-wall myocardium, and subepicardium, with 42.6% vs 45.6% vs 50% (P = 0.608). However, among 50 VT corridor sites from epicardial mapping, the adjacent LM tended to distribute in the subepicardium, with 46% vs 50% vs 74% (both P < 0.05) (Table 4).

DISCUSSION

The main findings of this study are as follows: 1) LM exhibits different intracardiac electrogram features from fibrosis; 2) LM proximity is closely associated with electrophysiologically proven VT re-entry corridors; 3) LM potentially facilitates VT by increasing impedance adjacent to the slowly conductive corridors, thus isolating and stabilizing current propagation along the corridor; and 4) VT corridor sites (ie, isthmus and exit sites of VT re-entry) highly colocalize with LM distribution.

Myocardial LM is thought to result from the replacement of fibrocytes by adipocytes following myocardial infarction.^4,5 In the current study, similar to results reported by Baroldi et al,⁶ we show that fibrosis in the setting of NICM also seems to trigger myocytes or fibrotic cells transformation to adipocytes. In postinfarct intramyocardial hemorrhage after coronary perfusion of large animal models, Cokic et al¹⁵ reported colocalization of LM, iron deposition and massive accumulation of foam cells, and suggested that intramyocardial hemorrhage might be an important cause of LM. The pathophysiology of LM in the setting of NICM has not yet been explored. In the chronic disease settings of hypertension, valvular heart disease, diabetes, and obesity, abnormal macrophage accumulation within the myocardium can induce intensive perivascular and interstitial fibrosis,¹⁶ as well as myocardial steatosis, both of which may act as precursors of LM.

To the best of our knowledge, this study is the first that reports the intracardiac electrogram features of myocardial LM and the association of its proximity with VT circuitry in patients with NICM. We show that surviving tissue near LM is electrophysiologically characterized by lower voltage, and higher impedance compared with the scar, similar to LM characteristics in patients with prior myocardial infarction.⁴ The corridors containing LM tend to have more myocardial mass than those containing only fibrosis but exhibit lower voltage and current, which means the former contains more severely impaired myocardium. It is important to note that LM is not conductive, but the surviving myocardial bundle traversing through or mixed with LM is conductive. A minimum amount of LM is needed to be detectable despite the high spatial resolution of 0.45 mm × 0.45 mm × 0.45 mm used in this study.

Corridor current amplitude varies more along non-VT corridors, expressed as larger SD of the corridor current, than that observed in VT corridors of the NICM patient in the present study. VT corridors contained much higher LM mass, exhibiting higher impedance than fibrosis. Thus, compared with fibrotic tissue, LM more effectively isolates VT corridors from neighboring tissue and likely decreases extracellular current from penetrating the high-impedance LM tissue and disseminating into the adjacent low-impedance fibrotic tissue, thereby contributing to stabilizing current propagation along VT corridors. In contrast, for those corridors near fibrosis, the fibrotic tissue can serve as a current sink. Thus, extracellular current from the latter corridor type tends to diffuse into the low-impedance fibrotic current sink, ultimately reducing intracellular current and the likelihood of propagation along the surviving corridor. Notably, current amplitude in non-VT corridors exhibited larger variation, which was different from our prior study reporting a general trend of current amplitude decline in non-VT corridors of postinfarct patients.⁴ In chronic ischemic cardiomyopathy, the LGE area from CMR consists of fibrosis and LM and compactly distributes along the culprit coronary artery territory. By contrast, in the NICM setting, fibrosis and LM are scattered and patchy, and we observed that the BZ volume was substantially larger than the dense LGE. This factor is potentially associated with our finding that non-VT corridors had significantly more connecting branches than VT corridors. These corridor branches likely affect the current propagation along the main corridors by injecting current into the main corridors or drawing current from the main corridors into the branches. Furthermore, additional branches increase the volume of the sink and worsen the source-sink mismatch when no LM is present to isolate branches from the main corridor.¹⁷ Consequently, myocardial LM is likely to mediate current propagation by decreasing extracellular current diffusion into the neighboring fibrosis and minimizing the impact of corridor branches. Thus, myocardial LM mitigates the volume of the sink and decreases source-sink mismatch. Our finding that VT corridor sites were highly colocalized with myocardial LM also supports the role of LM in NICM VT circuitry.

Importantly, we show that myocardial LM only constitutes a median of 0.2% to 1.7% per American Heart Association segment, whereas the median myocardial fibrosis content per patient reached 62.7% in patients with NICM referred for VT ablation.¹⁰ Pouliopoulos et al⁵ elucidated that myocardium adjacent to LM and fibrosis, compared with that adjacent to fibrosis alone, exhibits a higher level of myocardial discontinuity and cellular surface connexin 43 lateralization in a postinfarct ovine model. Given the lack of an animal model for NICM, cellular interactions between LM, fibrosis, and surviving myocardium are unknown in this setting. However, it is likely that the interaction of LM with adjacent fibrocytes and residual myocytes is complex, involving direct intercellular junctions, intercellular and extracellular current transmission and diffusion, which ultimately facilitate VT circuitry in NICM.

CLINICAL IMPLICATIONS.

These findings could improve the mechanistic understanding of VT circuitry in patients with NICM. Corridor proximity to LM is associated closely with corridor participation in VT and, therefore, may enhance the specificity of ablation targeting, tissue preservation, and maintenance of cardiac function in the long term. Most important, understanding the mechanism of current perpetuation within small corridors of viable tissue may provide therapeutic approaches, including targeting and mitigating cardiac adipogenesis, by developing novel strategies to increase the current sink and elevate the source-sink mismatch within scar myocardium.

STUDY LIMITATIONS.

Despite the prospective 2-center design, the study has several limitations. The results should be externally validated in a larger patient cohort. The current study provided no histologic evidence of myocardial LM. Our definition of continuous pixels of ≥1 mm³ with negative attenuation on myocardial CT images should be specific for LM identification.^4,18 Nevertheless, cardiac devices can cause artifacts which may exhibit negative attenuation and can be misclassified as LM. We excluded device artifacts, which were homogenously distributed along the direction of ICD leads. Despite best efforts, an estimated median registration error of 3.2 mm remained between endocardial mapping and images; however, this error is similar in magnitude to prior studies and nondifferential with regard to tissue characteristics.^11,19 A median registration error of approximately 4.0 mm remained on epicardial mapping. However, this registration error contains the thickness of epicardial adipose tissue and is similarly nondifferential regarding tissue characteristics.

The clinical usefulness of myocardial LM was not assessed prospectively in the current study. A prospective randomized trial to compare the safety and efficacy of LM guided vs standard VT ablation is warranted. Intracardiac points were acquired using point-by-point mapping. Multipolar catheters were not used for most cases in the present study. The application of multipolar catheters in future studies may refine the resolution of electrogram data. Impedance was measured by injection of alternating current from the catheter tip in a unipolar setup to 2 grounding patches, and the unipolar voltage was measured from the ablation catheter tip to the Wilson central terminal. Therefore, impedance and unipolar voltage were not measured using the same anode. However, as discussed in our previous study, although absolute measurements are affected, the relative change of current amplitude among different sites should be accurate.⁴ Nevertheless, future studies using an identical anode, and/or bipolar impedance and voltage measurements are planned and may refine these results. Due to the hemodynamic instability, activation mapping was only performed 45% in our cohort, and approximately a third of patients had identified VT isthmus sites. Finally, some mid-wall VT circuit sites were likely missed.

CONCLUSIONS

Corridors adjacent to LM tend to participate in VT circuits in patients with NICM. LM increases surrounding impedance and may enhance current propagation along VT corridors.

PERSPECTIVES.

COMPETENCY IN MEDICAL KNOWLEDGE:

LM proximity is spatially associated with electrophysiologically proven VT re-entry corridors in NICM. LM may facilitate VT by increasing impedance adjacent to the slowly conductive corridors.

TRANSLATIONAL OUTLOOK:

Corridor proximity to LM is associated with corridor participation in VT and, therefore, may enhance the specificity of ablation targeting. Mechanistic understanding of NICM VT circuitry may improve risk stratification and promote pharmaceutical or procedural strategies for eliminating or mitigating LM.

ABBREVIATIONS AND ACRONYMS

BZ

border zone

CMR

cardiac magnetic resonance

CT

computed tomography

EAM

electroanatomical mapping

HU

Hounsfield unit

ICD

implantable cardioverter-defibrillator

LGE

late gadolinium enhancement

LM

lipomatous metaplasia

NICM

nonischemic cardiomyopathy

VT

ventricular tachycardia