Introduction

Ventricular arrhythmias (VAs) in patients with structurally normal hearts are referred to as idiopathic. Frequent premature ventricular contractions (PVCs) are the most common form of idiopathic VAs, accounting for approximately 90% of them (1). However, PVCs may occasionally conceal underlying structural heart disease. The distinction between truly idiopathic PVCs and those related to structural heart disease is essential, because the former are considered benign and associated with similar risk of outcome events compared with the general population, whereas the latter are associated with increased cardiovascular morbidity and risk of sudden cardiac death (SCD) (2). Historically, diagnostic strategies including 12-lead electrocardiography (ECG), transthoracic echocardiography, exercise stress test, functional imaging, and assessment of coronary arteries by computed tomography coronary angiography or invasive coronary angiography have been implemented to rule out structural heart disease; however, these strategies may occasionally fail to identify subtle myocardial abnormalities (1,3). In the last 2 decades, cardiac magnetic resonance (CMR) imaging has emerged as a useful imaging technique in a broad spectrum of cardiac diseases, due to its ability to accurately and reproducibly assess biventricular volumes and function and to provide in vivo myocardial tissue characterization (4). However, whether CMR imaging can increase the ability to detect myocardial abnormalities in patients with frequent PVCs and a normal heart by routine diagnostic workup is unknown; similarly, little is known about its prognostic usefulness in this population. Therefore, the aim of the present study was to investigate the prevalence and prognostic significance of abnormalities found on CMR imaging in a large multicenter cohort of patients with apparently idiopathic frequent PVCs.

Methods

Study population

This was an observational CMR registry that included 518 patients with frequent (>1,000/24 h) PVCs and negative routine diagnostic workups. Patients were enrolled from January 2002 to December 2017 at 7 medical centers and were identified as those who underwent CMR imaging and who had frequent PVCs as the clinical indication. No pre-specified criteria were used to select patients for CMR, which was based on referring physician and/or patient preference. Patients who presented with nonsustained ventricular tachycardia (VT) were excluded. Negative routine diagnostic workup was defined on the basis of: 1) absence of systemic diseases and plasma electrolyte abnormalities; 2) normal 12-lead surface ECG; 3) normal 2-dimensional transthoracic echocardiography; and 4) absence of significant coronary artery disease demonstrated by maximal exercise stress test, functional imaging, computed tomography coronary angiography, or invasive coronary angiography. A review of all baseline ECGs and transthoracic echocardiographies was performed at each center by a single experienced operator (at least 5 years of clinical experience) blinded to clinical and outcome data. Details on the collection of clinical data are reported in the Supplemental Appendix. PVCs were defined as multifocal when ≥10% of the total PVCs on 24-h Holter monitoring differed from the dominant PVC morphology for at least 1 of the following criteria: opposite frontal axis (superior axis vs. inferior axis); difference in the bundle branch block pattern in lead V₁; or different precordial transition in case of same bundle branch block pattern in lead V₁. All patients underwent comprehensive CMR with late gadolinium enhancement (LGE) imaging for evaluation of myocardial necrosis and/or replacement fibrosis (Table 1). A de-identified database from consecutive patients with full Digital Imaging and Communications in Medicine (DICOM) datasets from the participating centers was used for data collection and analysis. Institutional review board approval was obtained at each center.

Outcomes

The study outcome consisted of a composite endpoint of malignant arrhythmic events including: 1) SCD, defined as unexpected death within 1 h of symptom onset; 2) resuscitated cardiac arrest due to ventricular fibrillation (VF) or hemodynamically unstable (i.e., causing severe hypotension with systolic blood pressure <90 mm Hg and syncope) VT; and 3) nonfatal episodes of VF or sustained VT that required appropriate implantable cardioverter-defibrillator (ICD) therapy.

Clinical follow-up

Patients were routinely evaluated at 3-month to 6-month intervals after CMR by clinic visits and phone calls to confirm the absence of arrhythmia symptoms. Medical records were reviewed to determine the occurrence of hospital admission due to cardiovascular causes and implantation of an ICD. Vital status and date and cause of death were determined by querying the respective social security death indexes. ICD interrogations were reviewed to assess ICD therapies among ICD recipients; ICD shocks or antitachycardia overdrive pacing were considered an appropriate therapy if triggered by VF or sustained VT.

Statistical analysis

Continuous variables were expressed as mean ± SD if normally distributed or median (25^th to 75^th percentile) if not normally distributed. All continuous variables were tested for normal distribution using the 1-sample Kolmogorov-Smirnov test. Categorical data were expressed as counts and percentages. Categorical variables were compared using the chi-square test or Fisher exact test when appropriate. Univariate and multivariable logistic regression analysis was performed to evaluate the relationship between the presence of abnormalities on CMR (including wall motion abnormalities, fat infiltration, and areas of LGE) and baseline covariates. Survival curves were generated by the Kaplan-Meier method and compared by the log-rank test. Two-tailed tests were considered statistically significant at the 0.05 level. All the analyses were performed using SPSS version 24.0, (IBM, Armonk, New York).

Results

Study population

Mean age was 44 ± 15 years, and 295 (57%) patients were men; 398 (77%) patients were symptomatic for palpitations, 35 (7%) for dizziness, 12 (2%) for dyspnea, and 8 (2%) for atypical chest pain, whereas 65 (13%) patients were asymptomatic (Table 2). Among asymptomatic subjects, PVCs were occasionally found during routine screening for sports eligibility in 48 (74%) subjects, on pre-operatory ECG in 12 (9%) subjects and documented by continuous ECG monitoring during surgery or other invasive procedures in the remaining 5 (8%) subjects. The median 24-h PVC burden was 16% (25^th to 75^th percentile: 10% to 20%) of the total beat count. In 126 (24%) patients, the dominant PVC had a right bundle branch block (RBBB) morphology. A left bundle branch block (LBBB) with an inferior axis was the most common morphology (n = 364; 70%), whereas a superior axis was more frequent among patients with a RBBB pattern compared with those with a LBBB pattern (n = 56 [44%] vs. n = 28 [7%]; p < 0.01). Multifocal PVCs were observed in 39 (8%) patients with a median number of PVC morphologies of 2 (25^th to 75^th percentile: 2 to 3). Multiple PVC morphologies were more frequently found in patients with a RBBB pattern compared with patients with a LBBB pattern (n = 28 [22%] vs. n = 11 [3%]; p < 0.01). Furthermore, 33 (85%) of the patients with multiple PVC morphologies had at least 1 PVC morphology with a RBBB pattern (Table 3).

CMR findings

All patients had biventricular volumes and function within normal 25^th to 75^th percentile. Overall, 85 (16%) patients had evidence of myocardial abnormalities (Table 4). Regional left ventricular wall motion abnormalities were observed in 27 (5%) patients. The most frequent location was represented by the basal mid-inferolateral wall in 14 (52%) patients, the inferior wall in 12 (44%) patients, and the basal mid-anterolateral wall in 10 (37%) patients. Regional right ventricular wall motion abnormalities were present in 26 (5%) patients, predominantly involving the basal mid-lateral wall in 16 (62%) patients and the right ventricular outflow tract in 8 (31%) patients. Of note, 3 (12%) of these patients met a minor CMR criterion for the diagnosis of arrhythmogenic right ventricular cardiomyopathy (5). Intramyocardial fat signal was observed in 9 (2%) patients in the left ventricle (with predominant involvement of the basal mid-segments of the anterolateral [n = 5] and inferolateral [n = 4] walls) and in 3 (1%) patients in the right ventricle (with predominant involvement of the lateral free wall). All the 3 patients with evidence of intramyocardial fat in the right ventricle also had regional wall motion abnormalities. A total of 82 (16%) patients had evidence of areas of LGE either in the left ventricle (n = 73), the right ventricle (n = 3), or both (n = 6). The median amount of LGE (percentage of left ventricular mass) was 5% (25^th to 75^th percentile: 3% to 11%) and was most commonly localized within the inferolateral wall in 38 (48%) patients, the basal mid-inferior wall in 31 (39%) patients, and the basal mid-anterolateral wall in 27 (34%) patients (Figure 1). In most patients with evidence of LGE, its distribution was either mid-myocardial in 25 (32%) patients, subepicardial in 24 (31%) patients, or both in 14 (18%) patients, whereas only 7 (9%) and 9 (11%) patients had evidence of subendocardial or transmural areas of LGE, respectively (all of them had no evidence of significant coronary artery disease on computed tomography coronary angiography or invasive coronary angiography). Among the 79 patients with evidence of left ventricular LGE, 23 (29%) showed a peculiar ring-like pattern defined as subepicardial and/or mid-myocardial LGE that involved at least 3 contiguous segments in the same slice (6).

Correlation of clinical characteristics, arrhythmia features, and imaging findings

Overall, the prevalence of myocardial abnormalities identified by CMR was higher among patients with multifocal PVCs compared with patients with PVCs of a single morphology (n = 27 [69%] vs. n = 58 [12%]; p < 0.01) and in those with RBBB patterns as the clinically dominant PVC morphology compared with a LBBB pattern (n = 64 [51%] vs. n = 21 [5%]; p < 0.01). Thirty-nine (70%) patients with a RBBB superior axis had evidence of abnormalities on CMR compared with 25 (36%) patients with a RBBB inferior axis: 5 (18%) of those with a LBBB superior axis and 16 (4%) of those with a LBBB inferior axis (p < 0.01) (Figure 2, Supplemental Table 1).

At multivariable analysis, age (odds ratio [OR]: 1.04; 95% confidence interval [CI]: 1.01 to 1.06; p < 0.01), male sex (OR: 4.28; 95% CI: 2.06 to 8.93; p < 0.01), family history of SCD and/or cardiomyopathy (OR: 3.61; 95% CI: 1.33 to 9.82; p = 0.01), multifocal PVCs (OR: 11.12; 95% CI: 4.35 to 28.46; p < 0.01), and a PVC pattern other than a LBBB inferior axis (OR: 14.11; 95% CI: 7.35 to 27.07; p < 0.01) were all independently correlated with the presence of abnormalities found on CMR (Table 5, Figures 3 and 4). Interestingly, 7 (34%) patients with a ring-like LGE pattern had a family history of SCD and/or cardiomyopathy, and all of them had PVCs with RBBB morphology compared with 6 (11%) patients and 38 (68%) patients with other LGE patterns (p = 0.03 and p < 0.01, respectively).

Long-term outcomes

During the study, 18 (4%) patients underwent primary prevention ICD implantation due to the presence of myocardial abnormalities found on CMR that were associated with either a history of unexplained syncope and/or family history of SCD (n = 15) or induction of hemodynamically unstable sustained VT during programmed ventricular stimulation (n = 3). Following CMR examination, 271 (53%) patients underwent catheter ablation, and programmed ventricular stimulation was performed in 95 of them (35%). Acute procedural success was achieved in 234 cases (86%). Sustained VT was induced in 8 cases (8%). Details on the electrophysiological study protocol, catheter ablation, and its results are presented in the Supplemental Appendix. After a median follow-up of 67 months (25^th to 75^th percentile: 43 to 94 months), 3 (1%) patients had SCD, 19 (4%) had a resuscitated cardiac arrest, and 4 (1%) patients experienced appropriate ICD shocks. In the 3 patients who died, SCD occurred during physical (n = 2) or emotional (n = 1) stress. The occurrence of the composite outcome was significantly higher among patients with abnormalities found on CMR (n = 25; 29%) compared with those without abnormalities (n = 1; 0.2%; p < 0.01) (Central Illustration, Figure 5). A single outcome event of resuscitated cardiac arrest due to fast (>200 beats/min) monomorphic VT occurred in the group of patients without abnormalities on CMR, whereas among those with evidence of LGE, the composite outcome occurred in 13 (57%) patients with a ring-like LGE pattern compared with 11 (20%) of patients with other LGE patterns (p < 0.01). All 3 cases of SCD occurred in the group of patients with a ring-like LGE pattern. The incidence of the composite outcome was significantly higher among patients with inducible VT compared with those with noninducible VT (38% vs. 10%; p = 0.03), whereas it was not affected by successful ablation (Supplemental Figure 1). The prognostic relevance of CMR was confirmed among patients with baseline high-risk features are defined by at least 1 of the following between PVCs with a non-LBBB inferior axis morphology, multifocal PVCs, family history of SCD and/or cardiomyopathy, and history of unexplained syncope (Supplemental Figure 2).

Discussion

The present study documents the prevalence and prognostic significance of myocardial abnormalities identified by CMR imaging and missed by routine diagnostic workup (including ECG, echocardiography, and assessment of myocardial ischemia and/or coronary anatomy) in patients with frequent PVCs. The main findings are as follows: 1) up to 16% of patients with apparently idiopathic PVCs present with an underlying structural heart disease, characterized in most of the cases by areas of left ventricular scar of nonischemic etiology; 2) older age, male sex, family history of SCD and/or cardiomyopathy are all clinical features associated with the presence of concealed myocardial abnormalities; 3) presence of multifocal PVCs and a non-LBBB inferior axis pattern are strongly correlated with the presence of abnormalities found on CMR; and 4) the presence of abnormalities on CMR is associated with malignant arrhythmic events during follow-up.

Abnormalities on CMR in patients with apparently idiopathic PVCs

Previous observations demonstrated that in patients presenting with sustained VT/VF, a diagnostic workup limited to the assessment of ECG, echocardiography, and coronary anatomy was insufficient because clinically relevant myocardial abnormalities could be missed (7). Myocardial tissue characterization by CMR imaging could conversely identify myocardial abnormalities in more than one-half of survivors of sudden cardiac arrest who had an inconclusive diagnosis and in up to 38% of patients who presented with nonsustained or sustained VT and had an otherwise negative diagnostic workup (3,7–11). In contrast, limited knowledge is available regarding the clinical usefulness of CMR imaging in patients who present only with frequent PVCs and no underlying structural heart disease on the basis of routine investigations. Such patients represent a cohort historically considered at low risk of cardiovascular events or SCD, because previous longitudinal studies have demonstrated little or no difference in outcomes from the general population (12). So far, only 1 large-scale study has specifically addressed the issue of concealed structural abnormalities detected by magnetic resonance imaging in patients with frequent PVCs and otherwise unremarkable diagnostic workup (including ECG and transthoracic echocardiography), but it included only patients with PVCs with a LBBB pattern and an inferior axis (13).

In the present study, we included a large cohort of patients with frequent PVCs and normal ECG and transthoracic echocardiographic findings, regardless of the PVC morphology, and found that the prevalence of myocardial abnormalities identified by CMR was non-negligible because almost 1 of 6 patients had myocardial structural abnormalities identified by CMR. This demonstrated the suboptimal diagnostic performance of routine diagnostic tests and the clinical usefulness of CMR for the detection of possible concealed cardiomyopathic substrates. A pool of baseline variables (including age, male sex, family history of SCD and/or cardiomyopathy, and PVC morphology) was independently correlated with the presence of abnormalities found on CMR. In particular, PVC morphology other than the LBBB inferior axis (i.e., non−outflow tract-related) and the presence of multifocal PVCs were both independent predictors of concealed myocardial abnormalities. Previous studies linked multifocal PVCs and VAs of the RBBB pattern with the presence of LGE on magnetic resonance imaging and worse clinical outcomes, but those studies included a heterogeneous group of patients with nonsustained and even sustained VT, known structural heart disease, and/or baseline reduced left ventricular ejection fraction, which were potential confounding factors (3,14,15). We observed abnormalities on CMR in up to 70% of patients with an RBBB superior axis pattern or multifocal PVCs (Figure 2), with 2% of patients having evidence of both myocardial fatty replacement and nonischemic scarring of the left ventricle. This possibly reflected a left dominant form of arrhythmogenic cardiomyopathy (16). Also, 10% of the patients had evidence of isolated subepicardial and/or mid-myocardial scar in the inferior or inferolateral left ventricular walls (Figure 6), which was possibly related to previous myocarditis or an early cardiomyopathic process (17,18). A recent postmortem study of young individuals with SCD documented an association between nonischemic myocardial fibrosis and known pathogenic gene variants linked with arrhythmogenic right ventricular cardiomyopathy, hypertrophic cardiomyopathy, and dilated cardiomyopathy (Figure 7) (18). Some data suggested that the severity of the hemodynamic impairment related to the PVC frequency, site of origin, and duration of exposure was related to the development of fibrosis. However, when fibrosis is the result of persistent exposure to a high burden of PVCs or precedes PVCs and is possibly the cause of PVC remains to be elucidated (19).

Prognostic value of CMR imaging in patients with apparently idiopathic PVCs

The diagnostic refinement of patients presenting with apparently idiopathic PVCs provided by CMR imaging has important practical consequences in terms of risk stratification. During a median follow-up of 67 months, we observed a 5% (0.9%/year) incidence of malignant arrhythmic events. Even if the higher incidence of the composite outcome compared with population-based studies (including subjects with frequent PVCs) might have reflected the selected nature of our study population, it is important to note that all but 1 event occurred in the group of patients with abnormalities found on CMR, with almost one-third of them experiencing the composite outcome (20). These findings demonstrated the clinical relevance of CMR imaging for risk stratification of subjects who were otherwise considered at low risk based on routine diagnostic investigations. Notably, all 3 observed SCD events occurred during physical or emotional stress, highlighting the arrhythmic risk of strenuous physical activity in subjects with frequent PVCs and underlying structural heart disease.

Clinical implications

Obtaining CMR imaging in all patients with frequent PVCs raises several issues, including lack of widespread availability of CMR, direct and indirect costs, and sustainability for health care systems. According to our results, few baseline clinical and ECG characteristics (i.e., age, male sex, family history of SCD and/or cardiomyopathy, and PVC morphology) could be proposed as “red flags” to identify those subjects who may have concealed cardiomyopathic substrates, and therefore, get the higher benefit from a diagnostic workup inclusive of CMR imaging. Patients with abnormalities on CMR represent a subgroup at high risk of life-threatening arrhythmic events who deserve proper medical attention and close clinical monitoring. The ultimate goal of risk stratification is to start appropriate interventions to improve the outcome of those subjects deemed at risk; avoidance of strenuous physical activity may be advised, considering its link with arrhythmic risk in the presence of structural heart disease (21). Further risk stratification with primary prevention ICD implantation may also be considered in selected cases, eventually tailored by VT inducibility on electrophysiologic study (15). However, the implementation of such interventions in routine clinical practice need further validation by large randomized clinical trials with long-term follow-up.

Study limitations

This was a multicenter study that involved tertiary referral centers for the management of VT. As such, the characteristics of the patient population might have been affected by selection and/or referral bias. This potentially explained the higher prevalence of some clinical features, such as family history of cardiomyopathy and/or SCD, as well as the higher incidence of the composite outcome compared with other unselected populations with PVC. The selection of patients to refer to CMR was at the discretion of the referring physician; thus, the lack of standardization might have introduced unmeasured confounders that led to the overestimation of the prevalence of abnormalities on CMR in our sample. The CMR findings were not blinded and might have influenced the decision to implant an ICD, which affected the likelihood of such patients to have VAs detected. However, this had a negligible impact on the outcomes recorded because all 4 patients who experienced ICD appropriate interventions had an ICD shock triggered by very fast (≥ 250 beats/min) VT or VF that would have been associated with SCD in the absence of an ICD. Electrophysiological study was not routinely performed due to a low prevalence of refractory PVCs; consequently, a rigorous relation between the origin of VAs and the site of myocardial abnormalities detected by CMR could not be established. T1 mapping, which could allow detection of interstitial fibrosis that possibly represented an arrhythmogenic substrate among those patients without detectable LGE, was not performed. A diagnosis of underlying cardiac disease in patients with abnormalities found on CMR could only be hypothesized because endomyocardial biopsy was performed in none of the patients; furthermore, genetic testing was not systematically performed. Finally, due to the relatively low prevalence of outcome events during follow-up, further larger studies are needed to more accurately evaluate the prognostic usefulness of finding abnormalities on CMR as predictors of cardiac events.

Conclusions

Patients with apparently idiopathic PVCs present concealed myocardial abnormalities detectable by CMR in 16% of cases. Clinical characteristics such as age, male sex, family history of SCD and/or cardiomyopathy, as well as arrhythmia features like multifocal PVCs and an ECG pattern other than LBBB inferior axis, are associated with the presence of abnormalities found on CMR. The presence of multifocal PVCs and myocardial abnormalities identified by CMR are both independent predictors of worse long-term outcome.

Footnotes