From 2012 to 2014, patients newly diagnosed with LVNC or DCM in 9 French centers for inherited cardiac diseases were prospectively included. Inclusion criteria: all patients above the age of 18 who were newly diagnosed (<6 months) with isolated LVNC or idiopathic DCM by TTE ± MRI more than 1 month after an episode of acute heart failure. We applied the Jenni criteria for LVNC selection (2): the presence of left ventricular (LV) trabeculations and deep recesses in communication with the LV cavity, as well as a ratio of NC/C > 2 in systole. The left ventricular ejection fraction was systematically measured using the Simpson biplane rule. Assessment of left ventricular longitudinal function by echocardiographic global longitudinal strain (GLS) was also reported when available. Cardiac MRI was also performed in 70 (71%) of the 98 patients with LVNC in our series, with an NC/C ratio of >2.3 in diastole as the recommended threshold for the diagnosis of LVNC using this technique (14). However, as previously reported (12), for the purpose of this study, only the echocardiographic criteria were used as inclusion criteria, with MRI confirmation in most cases. Finally, only patients with a validated diagnosis of LVNC confirmed were included. DCM was defined according to the ESC Working Group on Myocardial and Pericardial Diseases report (1) that accounts for LV or biventricular systolic dysfunction and dilatation that are not explained by abnormal loading conditions or coronary artery disease. All TTE results were independently validated by a core lab centralized review before inclusion. Underage patients, prevalent cases, valvular, ischemic, or congenital heart disease were excluded. This study complies with the Declaration of Helsinki and was approved by our institutional review board. Written informed consent was obtained from all patients. Participating centers managed the approvals of their ethics committees or Institutional Review Boards, according to local regulations.

Patients with LVNC also underwent a genetic analysis, as an ancillary study of the main registry (12).

All patients underwent a comprehensive baseline clinical and echocardiographic study with core lab analysis and genetic analysis. In addition to standard measurements, echocardiographic studies focused on trabecular quantification, extent, and localization. All TTE examinations were anonymized and transmitted to the main investigation center (Cardiology Department of the Timone Hospital, Marseille). A specific, unique code was used to identify each of the centers and patients. Demographics, comorbidities, and laboratory data were collected at baseline and at 1 and 2 years of follow-up. All patients were subjected to a physical examination and electrocardiography at the time of enrollment, which was repeated after 1 and 2 years. The scheduled end of the study visit was the 2-year follow-up visit for surviving patients. For deceased or transplanted ones, the last visit before the event was considered the end of the study visit. Data about the type of event and date of occurrence were obtained from patients' or clinicians' reports. A 5-year follow-up was also obtained through phone calls or contact with the patient or his practitioner.

The primary endpoint was the occurrence of cardiac death, heart transplantation, or hospitalization for cardiovascular complications (hemodynamic, rhythmic, or embolic) during follow-up. An initial follow-up was scheduled at 2 years, and then completed by a final evaluation at 5 years. Hemodynamic events were defined as the occurrence of pulmonary edema or hospitalization for acute heart failure. Embolic events include the occurrence of a stroke or of clinically detectable embolism in any territory requiring hospitalization. Rhythmic events included sustained ventricular tachycardia, syncope, or ICD-appropriate shock, as well as permanent or paroxysmal atrial fibrillation. Implantation of a pacemaker or defibrillator was also collected.

Secondary endpoints were the occurrence of any of the events that were comprised in the primary outcome.

To reduce the influence of LVEF on prognosis, a specific substudy was performed, including only patients with LVEF <45%, following the recent definition of DCM given by a position statement of the ESC working group on myocardial and pericardial diseases (15) and the expert consensus document from the European Association of Cardiovascular Imaging (11). Figure 1 summarizes the flow chart of the registry.

Continuous variables were expressed as mean and standard deviation and were compared using the Mann–Whitney nonparametric U test or Student's t-test, depending on the application conditions. Categorical variables were described by their number and percentage and were compared using the chi-squared test or the Fisher test, depending on the application conditions. Event-free survival distributions were estimated using the Kaplan–Meier method. The log-rank test was used to compare survival distributions for categorical variables, such as the cardiomyopathy variable. The effect of continuous variables on the risk of an event occurring was estimated and tested using a Cox model. Multivariate analyses were performed using the Cox model. Inclusion of variables in the fitted model was carried out according to the following criteria: p < 0.20 in univariate analysis as the threshold for entering variables, number of missing values <10% for these variables, effective presence in both categories of a bivariate variable >10%, and according to their clinical relevance. The final models selected were those with the lowest Akaike information criteria (AIC). Tests were performed in bilateral situations and were considered statistically significant for p < 0.05. Statistical analysis was performed using R software (version 3.6.2) and RStudio (version 1.2.5).

Between January 2012 and February 2014, a total of 98 LVNC and 65 DCM cases were prospectively included. The clinical and echocardiographic characteristics of the study population are summarized in Table 1. Age at diagnosis was similar between groups. No significant difference was found over symptoms except for the occurrence of syncope, which was more frequently reported in LVNC cases (p = 0.05). DCM patients presented with more severe LV dysfunction, including higher end-diastolic and end-systolic volumes and lower LVEF (43.3% [15–71] vs. 35.95% [15–55.7], p = 0.001). Concerning the LVNC group, mutations were identified in 47 (48%) patients, including sarcomeric mutations in 23 (48.9%) (10 TTN, 3 MYH6, 3 MYH7, 2 MYBPC3, 3 ACTC1, and 2 other sarcomeric genes) and ion-channel mutations in 15 (32% of mutated patients) (8 patients with mutations in HCN4 and 7 in RYR2).

The mean follow-up was 18.2 months for patients with LVNC and 17.8 months for patients with DCM, with 8 and 10 patients lost to follow-up, respectively. At 2-year follow-up, the primary endpoint was reached in 22 (22.4%) LVNC and 12 (17.9%) DCM cases, p = 0. 57. The mean 5-year follow-up was 73.69 months. The Kaplan–Meier curve in Figure 2 illustrates that the two groups' event-free survival was similar at a 5-year follow-up.

To compare patients with similar LVEF, we compared the prognosis of patients with LVEF <45% in both groups, including 56 patients with LVNC and 49 patients with DCM. No significant difference was observed in demographics or TTE parameters between the two groups (Table 2).

At 5-year follow-up, the primary endpoint occurred in 33 (58.9%) patients with LVNC and 18 (36.7%) patients with DCM (p = 0.02) (Table 3). The Kaplan–Meier curve in Figure 3 illustrates the 2 groups' event-free survival.

When looking at the occurrence of the primary outcome components taken separately (Table 3), hospitalization for heart failure (18 [32.14%] vs. 5 [10.20%], p = 0.035) and heart transplantation were much more frequent in the LVNC group at a 5-year follow-up than in the DCM group. Rhythmic complications were slightly more frequent in the LVNC group, but embolic events and cardiovascular death incidences were similar. Intracardiac defibrillator implantation was performed in 18 (32.1%) patients with LVNC and 10 (20.4%) patients with DCM. The number of patients experiencing more than 1 cardiovascular event during follow-up was also higher in patients with LVNC (Table 3).

Patients with LVNC and LVEF ≤ 45% (n = 56) were compared with patients with LVNC and LVEF > 45% (n = 42). Global longitudinal strain (GLS) values were available in 30 of 42 patients with LVNC and LVEF > 45%. GLS was relatively low in these patients (−18.1 [−22 to −13.2%], despite normal or near-normal LVEF (59.5 [47–68.4%]). Events were very rare in the latter group during the 5-year follow-up, with primary endpoints occurring in only four (9.5%) patients, including two hospitalizations for heart failure, three rhythmic complications, and no embolic events (Table 3, Figure 4).

Patients with LVNC and left ventricular dysfunction (LVEF <45%) present with a significantly worse prognosis at a 5-year follow-up as compared with matched patients with DCM and similar LV dysfunction.

Notably, heart failure events are more frequent in patients with LVNC than in those with DCM, while embolic events are not observed.

Conversely, patients with LVNC and LVEF >45% have very few events at 5 years.

The prognosis of LVNC has always been debated. Initially considered as a rare cause of heart disease (<1%) affecting young patients, LVNC natural course accounted for early and frequent cardiovascular events (7, 8). Initial cohorts incriminated LVNC for up to 38% of long-term mortality. Conversely, other studies reported a higher prevalence around 3–4% (16, 17) and described LVNC as a subgroup of cardiomyopathies with no prognostic difference (10, 18, 19). The diversity of prognosis and phenotypes has been assigned to the genetic background being itself widely heterogeneous and to various definitions that hamper comparisons between cohorts. The most recent studies are again contradictory. Aung et al.'s meta-analysis over 3 years of follow-up (19) and a recent Spanish retrospective study over 5 years of follow-up compared cardiovascular mortality between LVNC and DCM and argued both for equivalent prognosis (20). However, both studies suffer from several limitations; the most important being their retrospective nature. In addition, the meta-analysis was limited by variable LVNC definitions and heterogeneous primary endpoints when the Spanish study was limited by a small population, especially a few controls with only 13 DCM. More recently, Vaidya et al. compared the long-term survival of patients with LVNC with expected survival of age- and sex-matched US population and found that patients with LVNC have a reduced survival rate. However, this study was retrospective and did not compare the prognosis of patients between LVNC and DCM (21). Sedaghat-Hamedani et al. (22) compared an age-matched population of LVNC and DCM. Among 65 LVNC and 247 DCM cases, cardiovascular events were significantly more frequent in patients with LVNC than in those with DCM (p = 0.002, HR = 2.481). Unfortunately, in these series, patients were matched only on age and not on LVEF.

Our study is the largest prospective cohort comparing LVNC to DCM. All centers followed the European Society of Cardiology's current recommendations, and patient care was conducted accordingly. Although the occurrence of cardiovascular death was similar in LVNC and DCM cases, the primary endpoint, including the occurrence of cardiac death, heart transplantation, or hospitalization for cardiovascular complications (hemodynamic, rhythmic, or embolic), was more frequent at 5-year follow-up in the LVNC group.

Overall survival depends on the occurrence of cardiovascular events with different impacts of hemodynamic, rhythmic, and embolic factors.

Our study suffers from some limitations. First, the diagnosis of LNVC is frequently difficult, and the differentiation between LVNC and DCM may be difficult in some patients (5, 6). However, selection criteria were very strict in this study, with validation by a core lab specialized in the echocardiographic diagnosis of LVNC (6, 12). Second, the number of events was relatively low, particularly in patients with LVEF >45%, but the combined endpoint was clearly more frequent in patients with LVNC who had LV dysfunction as compared with DCM patients. Third, cardiac MRI was not performed in all patients in our series, and the predictive value of late gadolinium enhancement and fibrosis detected by MRI could not be evaluated (29). In addition, although LVEF was calculated in all patients, GLS was not available in all patients. Interestingly, however, when measured in patients with normal or near-normal LVEF, GLS was relatively low, possibly indicating the presence of mild left ventricular longitudinal dysfunction in these patients (30). Finally, and more importantly, this study was the first prospective one to allow a head-to-head comparison of patients with LVNC and DCM, with similar age and ejection fraction.