Non-invasive identification of slow conducting anatomical isthmuses in patients with tetralogy of Fallot by 3D late gadolinium enhancement cardiovascular magnetic resonance

Abstract Funding Acknowledgements Type of funding sources: None. Background Patients with repaired tetralogy of Fallot (rTOF) remain at risk of sudden cardiac death due to reentrant sustained monomorphic ventricular tachycardia (SMVT). Slow conducting anatomical isthmuses (SCAI), in particular SCAI3 at the outlet septum, bordered by the pulmonary annulus and the ventricular septal defect patch, are the dominant substrate for SMVT. Electroanatomical mapping (EAM) is the invasive gold standard to identify SCAIs, and transection of SCAI by catheter ablation has been correlated with favorable long-term outcome. Non-invasive identification of SCAI for risk stratification and treatment planning is needed but has not been established yet. Three-dimensional (3D) late gadolinium-enhanced (LGE) cardiovascular magnetic resonance (CMR) facilitates accurate visualization of morphologically complex hearts with high-spatial resolution. Objective The study thought to determine whether 3D LGE-CMR can identify SCAIs. Methods Consecutive patients with rTOF who underwent right ventricular (RV) EAM and 3D LGE-CMR were included. LGE-CMR-derived 3D RV reconstructions were created (ADAS-3D) and merged with 3D RV EAM data. Mapping points were superimposed on the CMR-derived 3D reconstruction allowing for direct comparison of EAM data and local signal intensity (SI). The optimal SI cut-off to identify low bipolar voltage (LBV, BV<1.76mV) was determined by receiver operating characteristic carve. An abnormal AI on LGE-CMR was defined as AI with continuous SI above the obtained cut-off connecting AI borders. Results Forty-eight rTOF patients (34±16 years, 58% male) were included. At EAM, 21 patients had normal AI, and 20 and 7 had a SCAI (<0.5m/s) or blocked AI, which was AI3 in all. Patients with SCAI showed low BV of AI3 (median 0.7 [range 0.25-2.59] mV). In 11 patients, 14 SMVTs could be induced, all related to SCAI3. A total of 9240 points were analyzed, showing a significant correlation between BV and SI (R=0.4, P<0.001). The optimal SI cut-off to identify LBV was 42% of the maximal SI (MSI) (AUC 0.80; sensitivity, 74%; specificity, 78%). Using this cut-off of MSI, a SCAI or blocked AI3 could be correctly identified by LGE-CMR in all 27 patients, and a normal AI3 could be correctly confirmed by LGE-CMR in 14/21 patients with normal EAM findings (Figure). The sensitivity and specificity of 3D LGE-CMR for identifying SCAI or blocked AI3 were 100% and 67%, respectively. Of note, among patients with normal EAM findings, those with abnormal AI3 on LGE-CMR had significantly lower BV of AI3 than those with normal AI3 on LGE-CMR (2.06 [Range, 1.62-2.60] vs. 3.53 [2.22-5.67] mV, P<0.01). Conclusion 3D LGE-CMR can identify SCAI with 100% sensitivity and may identify diseased AI3 even before critical conduction delay occurs. This technique may allow for non-invasive risk stratification of VT and can refine patient selection for invasive EAM.