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

BackgroundPatients with repaired tetralogy of Fallot (rTOF) remain at risk of sustained monomorphic ventricular tachycardia (SMVT) related to slow-conducting anatomical isthmuses (SCAI). Invasive electroanatomical mapping (EAM) is the only available method to identify SCAI (SCAIEAM). We aimed to determine rTOF-specific high signal intensity threshold values (HSIt) to identify abnormal myocardium by 3D late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) and assess the performance of LGE-CMR to non-invasively identify SCAIEAM. MethodsConsecutive rTOF patients who underwent right ventricular EAM (RV-EAM) and 3D LGE-CMR were included (2012-2021). A SCAIEAM was defined as an anatomical isthmus (AI) with conduction velocity (CV) <0.5 m/s. LGE-CMR-derived 3D RV reconstructions were merged with 3D RV-EAM data. The HSIt was determined based on the comparison of local bipolar voltages (BV) and the corresponding local SI using ROC analysis. An abnormal AI on LGE-CMR (Abnormal AICMR) was defined as AI showing continuous high SI (>HSIt) between anatomical boundaries. ResultsForty-eight rTOF patients (34{+/-}16 years, 58% male) were included. Of 107 AIs on EAM (AI1 and 3 in all, AI2 in 11), 78 were normal-conducting AIEAM (NCAIEAM), 22 were SCAIEAM (SCAIEAM2 in 2 and SCAIEAM3 in 20), and 7 were blocked AIEAM3. All 14 induced SMVTs were related to SCAIEAM3. A total of 9240 EAM points were analyzed. HSIt was 42% of the maximal SI (AUC 0.80; sensitivity, 74%; specificity, 78%). On 3D-CMR RV construction, all 29 SCAIEAM or Blocked AIEAM were identified as abnormal AICMR. Among the 78 NCAIEAM, 70 were normal AICMR and 8 were abnormal AICMR. The sensitivity and specificity of 3D LGE-CMR for identifying SCAIEAM or blocked AIEAM were 100% and 90% (29/29 and 70/78), respectively. Among patients with NCAIEAM3 (n=28), those with abnormal AICMR3 (n=6) had significantly lower BV and slower CV compared with those with normal AICMR3 (n=22) (BV, 1.91 [1.62-2.60] vs. 3.45 mV [2.22-5.67]; CV, 0.69 [0.62-0.81] vs, 0.95 m/s [0.82-1.09]; both P<0.01). Conclusion3D LGE-CMR can identify SCAI with excellent sensitivity and specificity and may identify diseased AI3 even before critical conduction delay occurs, which may enable non-invasive risk stratification of VT and may refine patient selection for invasive EAM. What is new?O_LIrTOF-specific high signal intensity threshold (HSIt) value on 3D LGE-CMR to identify abnormal myocardium was determined by direct comparison between 9240 superimposed 3D EAM points and corresponding local signal intensity on the 3D CMR-derived reconstruction. C_LIO_LIThe newly proposed method of CMR image analysis using the obtained HSIt showed an excellent interobserver agreement and could identify SCAI or blocked AI with 100% sensitivity and 90% specificity. C_LIO_LICompared to patients with NCAIEAM and normal AICMR (true negative CMR), those with NCAIEAM but abnormal AICMR (false positive CMR) had already significantly lower BV and CV on EAM. C_LI What are the clinical implications?O_LIThe newly proposed technique of 3D LGE-CMR image analysis may allow for non-invasive and serial risk stratification of VT in patients with rTOF and can refine patient selection for invasive EAM and concomitant ablation. C_LI