QF1-001 - Correlation of high-resolution 3D MR-derived gray zone with VT cycle length in a swine model of chronic infarction

Thursday, February 13, 2020

2:45 PM – 2:48 PM

Location: Sabal/Sago (Meeting Room in Exhibit Hall)

Background: Scar-related ventricular tachycardia (VT) is a major cause of death; with VT cycle length depending on the infarct size [1]. The VT substrate (a mixture of viable myocytes and collagen fibrils) facilitates the electrical waves to traverse scars and loop around them. Thus, accurate localization of the VT substrate and scar are critical prior to successful RF ablation therapy [2]. CMR can identify VT substrate in LGE images as a ‘grey zone’ (GZ) of intermediate signal intensity (SI) between healthy tissue and dense scar; however, the clinical spatial resolution is still limited [3]. Our goal is to detect GZ/scar in a swine model of chronic infarction using improved 3D CMR methods histologically validated, and to correlate the amount of MR-derived (GZ+scar) with VT cycle length.

Methods: Myocardial infarction was created in n=6 swine via 90min occlusion-reperfusion [4]. After 5weeks, the pigs underwent CMR on a 3T GE scanner. In all pigs, after a Gd-DTPA bolus (0.2mmol/kg), we employed conventional 3D LGE methods (TR=3.7ms, TE=1.6ms, optimized TI=225ms). Standard SI thresholds were used to derive GZ/scar areas based on 5SD [3]. In 3 pigs we also tested a fully-sampled version of a 3D multi-contrast late enhancement (MCLE) method, where 10 MCLE phases/cardiac cycle were acquired per volume and used to derive GZ/scar areas from T1*-SS (steady state) maps using a fuzzy-logic algorithm described in [6].

All pigs underwent then an x-ray guided electro-anatomical voltage mapping EAVM of the left ventricle (LV) and VT inducibility. For analysis, VT cycle length was correlated with the amount of MR-derived (GZ+scar) calculated relative to the LV volume. Select tissue slices were histologically stained with collagen-sensitive stain, to facilitate the correspondence of fibrosis to MR-derived tissue zones.

Results: Figures 1a-b show an example LGE and MCLE images acquired at the same level through a pig heart with an anterior-septal infarct, along with the resulting segmentation of tissue into: healthy myocardium, GZ, and scar. The CNR in MCLE images was increased by 2.7 and 1.9 times between infarct-blood and infarct-healthy tissue, compared to LGE images (1c). The pattern and location of MR-derived GZ and scar pixels matched with histologically-defined fibrosis in the peri-infarct and collagenous areas (1d) and a viable sub-endocardial layer, and with low bipolar voltages recorded by CARTO system (2a). There was a strong correlation (R²=0.84) between the amount of infarcted tissue and VT cycle length (2b), suggesting that a smaller amount of MR-derived (GZ+scar) is a predictor of fast VT.

Conclusion: We successfully demonstrated that 3D LGE and MCLE-derived GZ/scar zones derived at 1.4mm isotropic resolution have excellent correspondence to EAVM and VT wave characteristics, and to fibrosis in histological stains. Future work will focus on translating these 3D CMR methods in clinical routine, to accurately identify GZ and plan the ablation of VT substrate in patients with chronic infarcts.

Primary Author(s)

TE

Terenz Escartin, MSc, BSc

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.

Co-Author(s)

SS

Sheldon Singh, MD

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
JQ

John Quinto

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
JB

Jennifer G. Barry

Preclinical Coordinator
Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
PK

Philippa Krahn, BSc

Sunnybrook Research Institute & University of Toronto

Disclosure: Disclosure information not submitted.
XQ

Xiuling Qi, PhD

Research Associate
Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
MN

Matthew Ng, BSc

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
PL

Peter Lin, BSc

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
ML

Melissa Larsen

Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.
Graham A. Wright, PhD

Sr. Scientist
Sunnybrook Research Institute

Disclosure: Disclosure information not submitted.

Presenting Author(s)

Mihaela Pop, PhD, MSc, BSc

Scientist
Sunnybrook Research Institute

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.