Magnetic Resonance Left Ventricle Mass-Index/Fibrosis: Long-Term Predictors for Ventricular Arrhythmia in Hypertrophic Cardiomyopathy—A Retrospective Registry

Objective: We aimed to study the long-term association of LV mass index (LVMI) and myocardial fibrosis with ventricular arrhythmia (VA) in a population of patients with confirmed hypertrophic cardiomyopathy (HCM) using cardiac magnetic resonance imaging (CMR). Methods: We retrospectively analyzed the data in consecutive HCM patients confirmed on CMR referred to an HCM clinic between January 2008 and October 2018. Patients were followed up yearly following diagnosis. Baseline demographics, risk factors and clinical outcomes from cardiac monitoring and an implanted cardioverter defibrillator (ICD) were analyzed for association of LVMI and LV late gadolinium enhancement (LVLGE) with VA. Patients were then allocated to one of two groups according to the presence of VA (Group A) or absence of VA (Group B) during the follow-up period. The transthoracic echocardiogram (TTE) and CMR parameters were compared between the two groups. Results: A total of 247 patients with confirmed HCM (age 56.2 ± 16.6, male = 71%) were studied over the follow-up period of 7 ± 3.3 years (95% CI = 6.6–7.4 years). LVMI derived from CMR was higher in Group A (91.1 ± 28.1 g/m2 vs. 78.8 ± 28.3 g/m2, p = 0.003) when compared to Group B. LVLGE was higher in Group A (7.3 ± 6.3% vs. 4.7 ± 4.3%, p = 0.001) when compared to Group B. Multivariable Cox regression analysis showed LVMI (hazard ratio (HR) = 1.02, 95% CI = 1.001–1.03, p = 0.03) and LVLGE (HR = 1.04, 95% CI = 1.001–1.08, p = 0.04) to be independent predictors for VA. Receiver operative curves showed higher LVMI and LVLGE with a cut-off of 85 g/m2 and 6%, respectively, to be associated with VA. Conclusions: LVMI and LVLGE are strongly associated with VA over long-term follow-up. LVMI requires more thorough studies to consider it as a risk stratification tool in patients with HCM.


Introduction
Hypertrophic cardiomyopathy (HCM) is the most common inherited cardiac condition with an incidence of 1 in 500 births, characterized by cardiac hypertrophy, usually asymmetrical with the greatest involvement of the basal interventricular septum, with wall thickness ≥ 15 mm in adults or ≥13 mm without higher loading conditions [1][2][3][4][5][6]. Several HCM phenotypes have been identified and linked to >1400 mutations in 11 sarcomere protein genes with heterogenous presentation, diverse pathophysiology, and variable course [7].

•
To evaluate the association of LV MI and LV LGE with VT. • To compare differences in the ESC risk score using TTE and CMR.

Methods
Our study is a retrospective longitudinal observational study that included consecutive patients referred to outpatient HCM clinics of a single tertiary center with HCM diagnosed/confirmed diagnosis on CMR and followed between 2008 and 2018.
Local ethics committee approval was obtained from the Research and Development office of Nottingham University Hospitals NHS Trust, ID20-148C. The study was designed in 2018 and, at the time of conceptualization and design, no patient or public involvement was required or obtained for this retrospective study.
Baseline demographics and risk factors were retrieved from completed clinical records. HCM diagnostic criteria was LV WT ≥ 15 mm in adults or ≥13 mm in patients with a genetic mutation, after exclusion of secondary causes [3,4,11]. Initial and follow-up echocardiograms were performed in our tertiary center by an accredited BSE echo sonographer and reported by a level 3 cardiologist. LV WT was measured on TTE in the parasternal longand short-axis views at end diastole using a standard calibration scale [24]. ICDs were implanted as part of primary prevention guided by AHA and ESC guidelines before 2014 and the ESC risk score after 2014 [2,3,14]. Yearly follow-up for patients with echocardiogram and 48 h Holter monitoring occurred at each visit. LV ESC risk scores were calculated retrospectively for all patients using TTE and CMR, and categorized into low (<4%), moderate (4-6%) or high risk (≥6%) for SCD.
For primary outcome, patients were divided into two groups: Group A: patients with an incidence of VA: • NSVT: ≥3 consecutive ventricular beats ≥120 beats per minute captured on ECG, Holter monitoring or cardiac implantable electronic device (CIED) during follow-up. • Therapy: Appropriate anti-tachycardia pacing (ATP) or cardioversion/defibrillation provided by ICD or external cardioversion for sustained ventricular tachycardia (VT) or ventricular fibrillation (VF).
Group B: patients who had no VA detected during the follow-up and no ICD therapies.

Cardiac MRI
CMR examinations were performed using 1.5T scanners (Philips ACS-NT 1.5 T Gyroscan-Intera, Best, The Netherlands or Siemens Sonata 1.5 T, Erlangen, Germany) and a commercial cardiac coil. Electrocardiographic-gated, steady-state, free breath-hold sequential cines of 10 mm short-axis slices were acquired starting parallel to the atrioventricular ring and covering the entire ventricle. Late gadolinium enhancement images were acquired 15 min after the intravenous administration of 0.2 mmol/kg of gadolinium-DTPA (Magnevist, Schering; Berlin, Germany). A 2D segmented inversion-recovery sequence with breath-hold was acquired in the same views as the cine images.
LV volume, ejection fraction, mass and myocardial fibrosis were measured using standard volumetric and semi-automated techniques with commercially available software (Qmass MR version 6 1.6, Medis Medical Imaging Systems, The Netherlands), as shown in Figure 1. LV contours were outlined according to SCMR guidelines [25]. Trabeculae and papillary muscles were obviated from LV mass calculation [26]. LV WT was defined as the greatest dimension at any site within the LV wall ( Figure 2). To assess myocardial fibrosis (LV LGE ), all short-axis slices from base to apex were inspected visually to compare with areas of normal myocardium. Myocardial fibrosis was quantified at a grey-scale threshold of six standard deviations (SDs) above the mean signal intensity for normal myocardium ( Figure 3). The quantity of LV LGE was expressed as a percentage of the total LV myocardial mass [27]. The LV LGE analysis was performed on anonymized datasets twice by two experienced readers. Any discrepancies in analysis between the two readers were then adjudicated by a senior observer. To assess interobserver variability for the extent of LV LGE , 100 randomly selected studies were reanalyzed by the second reader. myocardial mass [27]. The LVLGE analysis was performed on anonymized datasets twice by two experienced readers. Any discrepancies in analysis between the two readers were then adjudicated by a senior observer. To assess interobserver variability for the extent of LVLGE, 100 randomly selected studies were reanalyzed by the second reader.

Statistics
All parametric continuous values were statistically analyzed using Student's t-test and presented as mean and SD. Non-parametric continuous values were analyzed using the Mann-Whitney test and presented as median and interquartile ranges. All categorical values were statistically analyzed using chi-square test or Pearson-Spearman test. Univariable Cox regression hazard proportional analysis was performed to evaluate hazard ratios (HR) and 95% confidence intervals (CI). Univariable results with p value < 0.1 were used in multivariable Cox regression analysis to establish if there was independent predictor of VA as defined in the methodology. Variance inflation factor was used to detect collinearity. myocardial mass [27]. The LVLGE analysis was performed on anonymized datasets twice by two experienced readers. Any discrepancies in analysis between the two readers were then adjudicated by a senior observer. To assess interobserver variability for the extent of LVLGE, 100 randomly selected studies were reanalyzed by the second reader.

Statistics
All parametric continuous values were statistically analyzed using Student's t-test and presented as mean and SD. Non-parametric continuous values were analyzed using the Mann-Whitney test and presented as median and interquartile ranges. All categorical values were statistically analyzed using chi-square test or Pearson-Spearman test. Univariable Cox regression hazard proportional analysis was performed to evaluate hazard ratios (HR) and 95% confidence intervals (CI). Univariable results with p value < 0.1 were used in multivariable Cox regression analysis to establish if there was independent predictor of VA as defined in the methodology. Variance inflation factor was used to detect collinearity. myocardial mass [27]. The LVLGE analysis was performed on anonymized datasets twice by two experienced readers. Any discrepancies in analysis between the two readers were then adjudicated by a senior observer. To assess interobserver variability for the extent of LVLGE, 100 randomly selected studies were reanalyzed by the second reader.

Statistics
All parametric continuous values were statistically analyzed using Student's t-test and presented as mean and SD. Non-parametric continuous values were analyzed using the Mann-Whitney test and presented as median and interquartile ranges. All categorical values were statistically analyzed using chi-square test or Pearson-Spearman test. Univariable Cox regression hazard proportional analysis was performed to evaluate hazard ratios (HR) and 95% confidence intervals (CI). Univariable results with p value < 0.1 were used in multivariable Cox regression analysis to establish if there was independent predictor of VA as defined in the methodology. Variance inflation factor was used to detect collinearity.

Statistics
All parametric continuous values were statistically analyzed using Student's t-test and presented as mean and SD. Non-parametric continuous values were analyzed using the Mann-Whitney test and presented as median and interquartile ranges. All categorical values were statistically analyzed using chi-square test or Pearson-Spearman test. Univariable Cox regression hazard proportional analysis was performed to evaluate hazard ratios (HR) and 95% confidence intervals (CI). Univariable results with p value < 0.1 were used in multivariable Cox regression analysis to establish if there was independent predictor of VA as defined in the methodology. Variance inflation factor was used to detect collinearity.
Receiver operative characteristic (ROC) curve analysis was used to define optimal cut-off values for LV MI and LV LGE as a test to predict VA. Negative and positive predictive values (NPV and PPV) were calculated for LV MI and LV LGE . These cut-off values were applied to measure incidence of VT using Kaplan-Meier curves measured using log rank test and HR. Correlation between LV MI and LV LGE was measured using linear regression and Bland-Altman plots. Statistical p values < 0.05 were considered significant. SPSS version 25 or higher (IBM corporation, Armonk, New York, NY, USA) were used for statistical analysis.

Predictors of VT
Cox regression univariable and multivariable analyses were performed to identify predictors for VA prior to the event in the HCM population, as shown in Table 2.

Discussion
Our study is one of the longest retrospective studies to observe HCM patients, and the main findings are as follows: 1. Higher LVMI is associated with VA and can be considered for risk stratification of SCD in HCM. 2. LVMI > 85 g/m 2 and LVLGE > 6% are associated with VA. 3. LVWT using TTE or CMR was only weakly correlated with LVMI on CMR. 4. LVMI and LVLGE were independent predictors of VA during follow-up. Similarly, there was significantly higher freedom from VA in patients with LVLGE < 6% compared to LVLGE > 6% (84.4% vs. 56.3%, p = 0.001, HR = 3.2, 95% CI = 1.57-6.56, p = 0.001) (Figure 9).

HCM risk score using TTE and CMR
American guidelines identified LV WT > 30 mm as a major risk factor for SCD and LV WT measured by TTE was incorporated into the European HCM Risk-SCD calculator [28]. TTE is the most common imaging modality used for diagnosis and risk stratification of HCM patients. However, LV MI and LV WT assessments are limited by asymmetric distribution of hypertrophy, echo window limitations resulting in underestimating or overestimating maximum LV WT , the inclusion of papillary muscles and right ventricular insertion into LV [19,[29][30][31]. Several studies reported considerable variation of LV WT assessed with TTE vs. MRI.Śpiewak et al. developed a simulation model comparing LV WT measured by TTE vs. CMR for risk stratification according to the European HCM Risk-SCD calculator. The discrepancy for which CMR measured LV WT translated to significant differences in the five-year risk of SCD [28].
CMR allows earlier and accurate diagnosis of HCM with early detection of myocardial fibrosis [9,14,30,31]. CMR manifests superior accuracy and reproducibility of LV WT and LV MI assessment compared to TTE [28,32,33], particularly when TTE imaging of LV is inadequate as supported by the British Society of Echocardiography and the current guidelines [3,9,14,30].
Another study showed the valuable role of using CMR in risk stratification of HCM patients. A study by Freitas et al., 2019, conducted a multicenter retrospective analysis of HCM. The study included 493 patients with median follow-up of 3.4 years. Their study showed that LV LGE identified and reclassified certain population underestimated with conventional risk scores [34].

Impact of LV WT , LV MI and myocardial fibrosis on cardiac events and prevention of SCD
Short-term and long-term studies have identified LV WT as an independent predictor for VA [4,35,36]. However, the value of LV WT to predict outcomes in patients with HCM is limited, with literature controversy. LVM offers more reliable representation of total LV hypertrophy compared to single wall thickness measurement, as shown in our study. Several studies highlighted the value of LV MI as an independent predictor for SCD. CMR is more accurate in the assessment of LV MI [28,32,33].
A recent retrospective study of 187 HCM patients by Dohy et al., 2021, demonstrated that CMR-derived LV MI is an independent predictor for a major event and myocardial fibrosis (LV LGE ) is a significant predictor for arrhythmia. Their patients were followed for an intermediate term (3.8 ± 2.4 years). The arrhythmia endpoint included malignant ventricular arrhythmia and appropriate ICD therapy. The incidence of death from all causes during follow-up was 10.7% (20/187) of patients. It is noted that their study population was younger (46.6 + 18.4 years) compared to our study (56.2 ± 16.3 years). Patients with ventricular arrhythmias had greater LV MI of 126.2 ± 56.5 g/m 2 and greater percentage of myocardial fibrosis of 13.1 ± 8.7% [33,37].
Myocardial fibrosis is another independent predictor for ventricular arrythmias, as shown in our study. Electrophysiological study of CMR LV LGE territories revealed a significant correlation between myocardial fibrosis and the abnormalities of catheter-mapped electrophysiological parameters in relation to the occurrence of malignant ventricular arrythmias. Ventricular arrythmia could be linked to conduction block created by myocardial fibrosis, and re-entry circuit created by residual non fibrotic myocardium [15,[38][39][40][41], with greater incidence of NSVT and ventricular ectopic with LV LGE . The risk of SCD is linearly related to LV LGE . Myocardial fibrosis > 15% was associated with a two-fold increased risk of SCD [16].

Implications of LV Mass/Fibrosis on Future Research in Hypertrophic Cardiomyopathy
There is a great interest in CMR-derived markers for risk assessment of patients with HCM. Our long-term study displayed significant association between LV MI and LV LGE and incidence of ventricular arrhythmias. Further investigations are required to assess the utility of adding CMR-derived markers for risk stratification of patients with HCM and improve identification of patients with HCM requiring ICDs for SCD prevention [42]. Myocardial biochemical changes have been demonstrated in some studies such as copper hemostasis.
Trientine demonstrated increased urinary copper excretion, with improvement in cardiac strain function along with a reduction in LV mass in this population [43]. LV LGE has limitations with sequences, heart rate and kidney disease. Hence, LV MI might be a more suitable, reproducible option without added software renderings and post-processing that is required for LV LGE for risk stratification. LV MI can be researched by conducting a randomized control study allocating moderate-risk patients with LV MI > 85 g/m 2 to either receive an ICD or a long-term continuous monitoring device such as an implantable loop recorder (ILR) for continuous risk assessment looking for NSVT as opposed to using extended ambulatory monitoring with poor diagnostic yields.

Limitations
There were limitations in our study related to retrospective data acquisition. The capture of VA using Holter monitoring is likely to result in underestimation compared to ICD monitoring, and the use of implantable loop recorders would give a more accurate capture of VA. Another limitation was the use of a combined endpoint e.g., NSVT, ATP and ICD shock. The last major limitation is that non-sustained VT does not necessarily translate to SCD over five-to ten-year follow-up, although this has not been extensively studied.

Conclusions
LV MI and myocardial fibrosis are strongly associated with ventricular arrhythmias over long-term follow-up of HCM patients. The utility of these CMR markers as risk stratification tools needs to be further investigated in a randomized control study.  Institutional Review Board Statement: Ethics approval was obtained and, due to the retrospective review, consent from individuals was waived. Individual consenting was not applicable. Research and Development office of Nottingham University Hospitals NHS Trust, ID20-148C.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.