In the present study, we sought to investigate the agreement between late gadolinium enhancement (LGE) in cardiovascular magnetic resonance (CMR) and electro-anatomical maps (EAM) of patients with non-ischemic dilated cardiomyopathy (NIDCM) and how it relates with the procedural outcome after catheter ablation of ventricular arrhythmias (VA). We identified 50 patients with NIDCM who underwent CMR and ablation for VA. LGE was detected in 16 patients (32%), mostly in those presenting with sustained VT (15 patients). Low-voltage areas (<1.5 mV) were observed in 23 patients (46%), in 7 patients (14%) without evidence of LGE. Using a threshold of 1.5 mV, a good and partially good agreement between the bipolar EAM and LGE-CMR was observed in only 4 (8%) and 9 (18%) patients, respectively. With further adjustments of EAM to match the LGE, we defined new cut-off limits of median 1.5 mV and 5 mV for bipolar and unipolar maps, respectively. Most VT exits were found in areas with LGE (12 out of 16 patients). VT exits were found in segments without LGE in 2 patients with unsuccessful ablation as well as in 2 patients with successful ablation, P=0.77. In patients with VT recurrence, the LGE volume was significantly larger than in those without recurrence: 12.2 ± 5.8% vs. 6.9 ± 3.4%; P=0.049. Myocardial heterogeneity provides the electrophysiological substrate of ventricular arrhythmias in patients with myocardial infarction. Fibrosis and reduction in the number of gap junctions of surviving myocytes allow the occurrence of re-entry (23). However, the relationship between fibrosis and VA is complex and involves not only fixed anatomical barriers but also functional blocks caused by differences in the fiber orientation, myocardial thickness mismatch or connexin downregulation (24-26). Studies involving EAM in patients with scar-related VT demonstrated that homogenization of the low-voltage areas with elimination of the signals showing abnormal amplitude and fractionation was associated with improved acute and long-term success rates (27). On the other hand, animal studies showed that CMR can be useful to characterize LV fibrosis. Moreover, the amount of LGE has been associated with inducibility of VT and is considered a powerful and independent predictor of adverse prognosis, especially in myocardial infarction patients (28-29). In contrast to ischemic cardiomyopathy, LGE is infrequently found in patients with NIDCM. A previous study of 399 patients with NIDCM demonstrated that LGE was detected in approximately one-fourth of the patients and was associated with a 9-fold increase of risk for SCD (30). In accordance with these data, we observed LGE in approximately one third of the patients, and most of them had a history of spontaneous sustained VT. In contrast to the VT patients who have frequently LGE, all patients with ventricular premature beats but without any sustained VT did not show any evidence of LGE in CMR. These observations support the general understanding that the presence of LGE identifies more advanced cardiomyopathy as well as a higher risk for more malignant ventricular arrhythmias. Although myocardial fibrosis is associated with a higher likelihood for VT occurrence, the absence of LGE in CMR does not completely eliminate the risk for VT. Some patients had sustained ventricular arrhythmias even without detectable scar in CMR, which suggests a poorer negative predictive value for the LGE. Although CMR imaging is currently considered the reference standard for the detection of LV scar, it has a limited spatial resolution in vivo. Therefore, minute scars as well as diffuse fibrosis that can still trigger VA may remain undetected. The alternative approach to detect myocardial scar is to characterize the electrical properties of the myocardium by using bipolar EAM in order to find low-voltage areas and late potentials that are markers of abnormal tissue. However, abnormal fragmentation and amplitudes below 1,5 mV are less frequently found in NIDCM in comparison to post-myocardial infarction patients. These findings illustrate the downsides of the EAM in NIDCM. Moreover, numerous animal and clinical studies underlined other technical drawbacks of the EAM that can influence the size and the characteristics of the low-voltage areas such as mapping electrode size and spacing, the angle of contact with the underlying tissue, wave-front direction (31-33). Recently, Betensky and al. analyzed the agreement between CMR and EAM in patients with NIDCM and found a significant discordance between both approaches in 36% of the patients. Using lower signal intensity threshold of 2 standard deviations they increased the CMR-EAM agreement up to almost 90% (34). In contrast to Betensky, who used a simplified approach analyzing only the septal to lateral disagreement, we choose to perform more precise analysis using the 17 segments AHA model of the LV. We found 23 out of 50 patients with low-voltage areas and 15 (71.4%) of them had sustained VT. Moreover only 16 (32%) patients with low-voltage had also LGE in the CMR. In our study the basal inferolateral, inferior and infero-septal segments were most frequently affected by LGE in contrast to the basal anterior and anteroseptal segments affected in the EAM. However, in the LGE positive patients, the best pace-mapping sites of the clinical VT coincided with areas of LGE. One possible explanation for the low correlation between EAM and LGE-CMR is the non-transmurality of the fibrosis in patients with NIDCM. A previous study in post-infarct patients demonstrated that median bipolar voltage <1.5 mV was only found in segments demonstrating ≥75% infarct transmurality (35). In a recently published article, Zeppenfeld et al. found that EAM voltages showed a linear relationship with the LV wall thickness and the amount of fibrosis in patients with non-ischemic DCM. However, no cutoff value for the voltage could be found to reliably delineate fibrotic areas in NIDCM (36). Regarding the quantification of the arrhythmogenic substrate, we could not find any correlation between the amount of LGE and the size of the low-voltage areas (endo- or epicardial), which can be explained by the impact of LGE transmurality as well as the sparse distribution of the LV fibrosis. In this regard, an advantage of the LGE-CMR is that it can visualize the presence of intramyocardial and epicardial scar which are not visible by endocardial EAM. The reason is that the bipolar EAM has narrower field of view and proved insensitive to delineate scar that lies deeper within the myocardium (37). Previously, Hutchinson et al. reported that by using a unipolar 8.27 mV threshold endocardial it was possible to identify epicardial bipolar low-voltage areas consistent with macroscopic scarring in patients with NIDCM and normal endocardial bipolar voltage (38). However, we found that the agreement between LGE and unipolar maps using this cutoff of 8.27 mV was poor. After adjusting the unipolar and bipolar threshold on the basis of CMR, the resulting median thresholds for the bipolar and unipolar low-voltage maps were 1.5 mV and 5 mV respectively, which are close to those observed in a previous study (37). 4.1 Conclusions LGE was observed in approximately one-third of the patients with dilated cardiomyopathy of non-ischemic origin and ventricular arrhythmias. LGE was seen mainly in patients with sustained VT. The agreement between the distribution or the extent of LGE and bipolar low-voltage areas was fairly poor. No particular cutoff values for bipolar and unipolar electro-anatomical maps could be found. On the other hand, most VT exits in patient with sustained VT were found in areas of LGE. The procedural success after VA ablation were related to LGE volume only.