Role of Myocardial Fibrosis in Hypertrophic Cardiomyopathy: A Systematic Review and Updated Meta-Analysis of Risk Markers for Sudden Death.

BACKGROUND: Hypertrophic cardiomyopathy (HCM) is associated with sudden death (SD). Myocardial fibrosis is reportedly correlated with SD.
OBJECTIVE: We performed a systematic review with meta-analysis, updating the risk markers (RMs) in HCM emphasizing myocardial fibrosis.
METHODS: We reviewed HCM studies that addressed severe arrhythmic outcomes and the certain RMs: SD family history, severe ventricular hypertrophy, unexplained syncope, non-sustained ventricular tachycardia (NSVT) on 24-hour Holter monitoring, abnormal blood pressure response to exercise (ABPRE), myocardial fibrosis and left ventricular outflow tract obstruction (LVOTO) in the MEDLINE, LILACS, and SciELO databases. We used relative risks (RRs) as an effect measure and random models for the analysis. The level of significance was set at p < 0.05.
RESULTS: Twenty-one studies were selected (14,901 patients aged 45 ± 16 years; men, 62.8%). Myocardial fibrosis was the major RISK MARKER (RR, 3.43; 95% CI, 1.95-6.03). The other RMs, except for LVOTO, were also predictors: SD family history (RR, 1.75; 95% CI, 1.39-2.20), severe ventricular hypertrophy (RR, 1.86; 95% CI, 1.26-2.74), unexplained syncope (RR, 2.27; 95% CI, 1.69-3.07), NSVT (RR, 2.79; 95% CI, 2.29-3.41), and ABPRE (RR, 1.53; 95% CI, 1.12-2.08).
CONCLUSIONS: We confirmed the association of myocardial fibrosis and other RMs with severe arrhythmic outcomes in HCM and emphasize the need for new prediction models in managing these patients.

Introduction

Hypertrophic cardiomyopathy (HCM) is an autosomal dominant disease characterized by ventricular hypertrophy in the absence of other conditions that cause heart overload.[1]-[3] It is the most common genetic disease, with a prevalence rate estimated at 1:500, affecting both men and women. Its presentation can vary from asymptomatic to more severe complications, such as sudden death (SD), which has an incidence rate of 1% per year. HCM is mainly responsible for SD in young and competitive athletes.[5]-[7] There is a discussion regarding how we should stratify SD and indicate implantable cardioverter-defibrillator (ICD) for the purpose of primary prevention of this disease. Strategies have been proposed to identify these patients, and classic risk markers (RMs), such as family history of SD, severe ventricular hypertrophy, unexplained syncope, non-sustained ventricular tachycardia (NSVT) on 24-hour Holter monitoring, and abnormal blood pressure response to exercise (ABPRE), have been evaluated in several clinical studies.[13] As such, the pathophysiology of SD in HCM is not fully understood. Some factors seem to be involved, including the development of myocardial fibrosis. Studies that investigated myocardial fibrosis using magnetic resonance imaging (MRI) have shown correlations with severe outcomes. In a recent study, Chan et al.[14] found that a percentage of fibrosis > 15% of the left ventricular mass was associated with a twofold increase in the risk of SD in patients considered initially at low risk.1 However, the detection of myocardial fibrosiB1 using cardiac MRI continues to generate discussions among experts and is now considered only a risk modifier, as evidenced by the American College of Cardiology Foundation / American Heart Association Task Force on Practice Guidelines.[1] The reassessment of RMs, considering the presence of myocardial fibrosis, is fundamental to improve risk stratification. We performed a systematic review and meta-analysis of observational studies that examined RMs in HCM, emphasizing the presence of fibrosis using cardiac MRI, to evaluate their statistical power in predicting SD.

Methods

Study design

Systematic review with meta-analysis of observational studies of the natural history of HCM that reported RMs of SD and severe arrhythmic outcomes.

Search strategy

The search used 3 databases - MEDLINE, LILACS and SciELO - contemplating prospective or retrospective studies conducted between 1980 and 2016, which analyzed the natural history of patients with HCM, regardless of sex or ethnicity. We used the PRISMA statement checklist to conduct the systematic review and meta-analysis. The detailed research adapted for each database was conducted using the following keywords of Medical Subject Heading (MeSH) and DECS: Cardiomyopathy, Hypertrophic, Familial OR "Cardiomyopathy, Hypertrophic" OR "cardiomyopathies" OR cardiomyopathy OR "risk factors" Death OR "defibrillators, implantable" OR "cardioverter defibrillator, implantable". Only the articles published in English, Portuguese, and Spanish were considered for the full-text review.

Selection criteria

We have included only observational studies (prospective or retrospective cohorts) that had a severe arrhythmic outcome equivalent to SD. The studies that also analyzed at least one of the following RMs were included: a) family history of SD, b) severe left ventricular hypertrophy, c) unexplained syncope, d) NSVT on 24-hour Holter monitoring, e) ABPRE, f) presence of left ventricular outflow tract obstruction (LVOTO), and g) presence of myocardial fibrosis on cardiac MRI. The exclusion criteria were (1) studies that were case reports or review articles, (2) studies that did not meet the previously described inclusion criteria, and (3) duplicate studies.

Definitions

The studies included in our meta-analysis often used variable concepts, but fit the definitions listed in Table 1.

Table 1

Definitions of outcomes and risk markers used in the meta-analysis

Severe arrhythmic outcomes	SD, aborted SD, documented sustained ventricular tachycardia, or appropriate shock in patients with ICD
Family history of SD	Family history of SD in the first-degree relatives of patients
Severe left ventricular hypertrophy	Ventricular thickness > 30 mm measured using echocardiography in any left ventricular segment
Unexplained syncope	A history of unexplained and transient loss of consciousness with spontaneous recovery
NSVT on 24-hour Holter monitoring	≥ 3 consecutive ventricular extrasystoles with heart rates of ≥ 120 bpm for < 30 seconds
LVOTO	Peak gradient of ≥ 30 mmHg in the left ventricular outflow tract detected on echocardiography
ABPRE	Increased (< 20 mmHg) or decreased (> 10 mmHg) systolic blood pressure with peak exercise
Myocardial fibrosis	Detection of late enhancement on MRI with gadolinium

SD: sudden death; ICD: implantable cardioverter-defibrillator; NSVT: non-sustained ventricular tachycardia; LVOTO: left ventricular outflow tract obstruction; ABPRE: abnormal blood pressure response to exercise; MRI: magnetic resonance imaging.

Data extraction

The eligibility (using inclusion and exclusion criteria) of each study was systematically analyzed by two reviewers (MIB and SAC), initially by reading the titles and abstracts. Selected articles were read and analyzed in full to assess their eligibility and methodological quality; all references were revised to identify additional studies. Differences in opinion between the two main reviewers were independently resolved by a third reviewer (DV). After this phase, the data were extracted. The information collected from each study included study design, number of patients, demographic data, follow-up, RMs for SD, and severe arrhythmic outcomes. Authors were contacted when any additional information was needed. There was no time restriction for the severe arrhythmic outcomes.

Statistical analysis

The statistical analysis used relative risks (RRs) as a measure of effect with 95% confidence intervals (CIs). The meta-analysis was performed using the DerSimonian and Laird method in case of heterogeneity and the Mantel-Haenszel method in case of homogeneity. The heterogeneity was analyzed using Cochran's Q and I[2] Higgins/Thompson tests. The risk of bias was tested using funnel plots and Egger's linear regression test. The software used for the analysis was R 3.4.1. The level of significance was set at p < 0.05.

Ethical and legal aspects

The study protocol was submitted to the Medical Ethics Committee of Pedro Ernesto University Hospital and received a final opinion on November 14, 2013.

Results

Search results

The search strategy identified 809 potentially relevant articles (Figure 1). After reading the titles and abstracts, 123 remained for the eligibility analysis. After detailed evaluation, 103 articles were excluded, and 1 article was added after reviewing the references. Thus, 21 observational studies were selected, including 14 prospective and 7 retrospective studies, comprising 14,901 patients (age, 45 ± 16 years; 62.8% males).[11]-[14],[17]-33 The main characteristics of the included studies are shown in Table 2.

Figure 1

Flowchart of the systematic review.

Table 2

Characteristics of the observational studies involving RMs of SD in HCM

Author / year of publication	Country	N. of patients	Age (years)	% males	Follow-up (months)	Severe arrhythmic outcomes
Elliott et al. 2006[17]	UK ;1988-2002	917	43	60.4	61	54
Elliott et al. 2000[18]	UK; 1988-1998	368	37	64.9	43.2	22
Gimeno et al. 2009[13]	UK; 1988 - 2004	1380	42	61.8	54	NI
Kofflard et al. 2003[19]	Netherlands; 1970-1999	225	41	57.7	96	20
Kofflard  et al. 1993[20]	Netherlands; 1970-1990	113	38	53.09	87.6	9
Maron B et al. 2007[21]	Multicentric; 1983-2005	383	41	62.9	44.4	51
Maron M et al. 2003[22]	USA and Italy; 1983- 2001	1101	45	59.4	75.6	71
Michaelides et al. 2009[23]	Greece; 1999-2001	81	42	70.3	63.6	8
Monserrat et al. 2003[12]	UK; 1988-2000	531	39	60.8	70	32
Rubinshtein et al. 2010[24]	USA; 2001-2007	424	55	59.1	43	8
Spirito et al. 2009[25]	USA and Italy; 1983-2005	1511	46	61.3	67.2	74
Spirito et al. 2000[11]	USA and Italy; 1983-1997	480	47	60	78	23
Syska et al. 2010[26]	Poland; 1996-2006	78	36.4	47.4	55.2	13
Chan et al. 2014[14]	USA and Italy: 2001-2010	1293	46	63	39.6	37
Spirito et al. 2014[27]	Multicentric; 1990-2009	653	44.4	70.5	63.6	24
Magnusson et al. 2016[28]	Sweden; 1995-2002	237	52	69.2	64.8	77
Klopotowski et al. 2015[29]	Poland; 2008-2013	328	45	58.5	37	14
Mahony et al. 2014[30]	Multicentric	3675	48	63.9	68.4	198
Debonmaire et al. 2015[31]	Netherlands and Belgium	195	52	61	68.4	26
Ismail et al. 201432	UK; 2000-2011	711	55	70.4	42	22
O´Hanlon et al. 201033	UK; 2000-2006	217	53.2	70.5	37.2	12

RMs: risk markers; SD: sudden death; HCM: hypertrophic cardiomyopathy; UK: United Kingdom; USA: United States of America; NI: Not informed.

Myocardial fibrosis

The systematic review selected 5 articles that correlated the presence of fibrosis on cardiac MRI with severe arrhythmic outcomes. One of them was excluded from the meta-analysis because all the patients who had events had myocardial fibrosis using the MRI, making it impossible to calculate the measure of effect.[24] Among the four remaining studies involving 2549 patients, the presence of fibrosis was correlated with events equivalent to SD in two studies;[14],[29] however, this only occurred in the univariate analysis of the other two.32,33 In the meta-analysis, we found a significant probability of severe arrhythmic outcomes in the presence of this variable (RR, 3.43; 95% CI, 1.95-6.03). We highlighted the absence of heterogeneity in the Forest plot and the RR, which was the highest among all the other markers evaluated (Figures 2A and 3). The funnel plot for myocardial fibrosis is shown in figure 2B.

Figure 2

A. Forest plot of myocardial fibrosis and relative risk of the severe arrhythmic outcomes. B. Funnel plot of myocardial fibrosis to evaluate for publication bias. TE: estimated treatment effect; seTE: standard error of treatment estimate; RR: relative risk; 95%CI: 95% confidence interval.

Figure 3

Forest plot of the risk markers and the relative risk of severe arrhythmic outcomes: A: unexplained syncope; B: severe ventricular hypertrophy; C: family history of sudden death; D: non-sustained ventricular tachycardia; E: abnormal blood pressure response to exercise; F: left ventricular outflow tract obstruction. TE: estimated treatment effect; seTE: standard error of treatment estimate; RR: relative risk; 95%CI: 95% confidence interval.

Meta-analysis of classic RMs

As shown in Figure 3, the following classic RMs demonstrated an association with the outcomes studied: family history of SD (13 studies - 9815 patients; RR, 1.75; 95% CI, 1.39-2.20); severe left ventricular hypertrophy (11 studies - 5501 patients; RR, 1.86; 95% CI, 1.26-2.74); unexplained syncope (12 studies - 10064 patients; RR, 2.27; 95% CI, 1.69-3.07); NSVT on 24-hour Holter monitoring (14 studies - 9421 patients; RR, 2.79; 95% CI, 2.29-3.41); and ABPRE (6 studies - 3061 patients; RR, 1.53; 95% CI, 1.12-2.08).

In our analysis, the only RISK MARKER that showed no correlation with serious arrhythmic outcomes was LVOTO (5 studies - 4762 patients; RR, 1.72; 95% CI, 0.97-3.02). This may have been influenced by the small number of studies involved in the combined analysis. It should be emphasized that the inclusion of this marker in the stratification of SD has always been a matter of debate.

When we evaluated the heterogeneity in all the RMs, we observed that only severe left ventricular hypertrophy was significant (I2 = 75%, p < 0.01), although we emphasize that the random effect model used in the meta-analysis already had mitigated this aspect. The Egger´s test also points to a publication bias for this RISK MARKER (p = 0.002)

We did not observe any publication bias by Egger's test or by the funnel plot for the other classic RMs.

Based on these results, it seems plausible that all classic RMs can still be used in SD stratification in HCM, except for LVOTO.

Discussion

This systematic review and meta-analysis shows the importance of a broad approach in SD risk stratification in patients with HCM, including myocardial fibrosis assessment.

The evaluation of patients with HCM may include multiple complementary examinations in an attempt to predict SD. Obviously, this has several effects, including economic burdening. Thus, knowing how to select the more important RMs is essential. Although primary prevention has been the object of research in several studies in the last decades, attempting to predict which patients with HCM have a higher risk of SD remains challenging. Therefore, systematic reviews and meta-analyses of such a controversial topic becomes important. The presence of the RMs studied here may define the need for ICD placement, considering that it is the only safe and effective tool in preventing SD.36 Some RMs have been reported to be more relevant, such as family history of SD, which was highlighted in the study by Dimitrow et al.,[10] however, its low positive predictive values is a limitation. In the event of syncope, it is only indicative of the risk when unexplained. Thus, all RMs have their limitations.

Strategies using the sum of classic RMs were not feasible. In the multicenter registry performed by Maron et al.[8] in patients with HCM who were treated with ICD placement, it was observed that 35% who received an appropriate shock had only 1 RM. These data were reinforced by a recent meta-analysis of patients with HCM and ICD who had 1.8 RISK MARKERS for SD on average, with a rate of 3.3% appropriate shocks per year. We also emphasize that the analysis did not include studies that had myocardial fibrosis as an RM.

Among the more recently studied markers aiming to establish correlations with an increased risk of SD in HCM, the most important was myocardial fibrosis. The mechanism suggested for this predisposition is that the presence of myocardial fibrosis could be a substrate for ventricular reentry areas. A classic study demonstrated that this finding correlates with the presence of NSVT in the 24-hour Holter monitoring. Shiozaki et al.39 in a recent national experience with 26 patients with HCM and ICD, assessed myocardial fibrosis by another method, the contrast-enhanced computed tomography, and found a higher rate of appropriate shocks in patients who had a fibrosis mass ≥ 18 g. Most of the studies that evaluated myocardial fibrosis in this population used cardiac MRI, and these were the experiences we analyzed with the focus on outcomes associated with MS.

Putting the results into context, this meta-analysis of observational studies reports a statistically significant association between myocardial fibrosis detected on cardiac MRI and outcomes equivalent to SD. Although we have assessed few articles, this is the most important finding in this study, showing the highest RR among all RMs studied with a very reliable CI. And even if the funnel plot has revealed discrete asymmetry suggesting a publication bias for myocardial fibrosis, it is important to note that the small number of articles does not allow one to conclude this assertion.

A meta-analysis published by Briasoulis et al.,40 addressing only myocardial fibrosis, found similar results. However, one article used in the analyses did not allow a precise calculation of an effect measure because the group without fibrosis did not have any event.[24] Our option was to remove it, because we understood this would compromise the statistical analysis. It stands out that this meta-analysis included the article of Klopotowski et al.[29] with 328 patients and updated other RMs.

We consider that this finding is of much clinical relevance, as the latest guidelines on the subject do not address the presence of fibrosis as an RM. In its latest document regarding the disease, the European Society of Cardiology based the indication of ICD placement on using a risk calculator (HCM-Risk SCD) created to provide more accurate stratifications.[2] Based on a cohort, the derived model used the parameters of age, maximum ventricular thickness, LVOTO, left atrial diameter, family history of SD, presence of NSVT, and unexplained syncope.[30] Subsequent studies showed conflicting results regarding the calculator., Perhaps, the fact that it does not assess fibrosis may be a limitation.

Regarding the other findings, we observed that all classic RMs correlated with the occurrence of the outcomes studied, except for LVOTO. In contrast to what has been observed in a previously published meta-analysis, our findings do not indicate that LVOTO may be associated with severe arrhythmic outcomes. This was probably because of the smaller number of patients used in our analysis and the inclusion of two recent studies of which results do not indicate the association between this marker and SD.

The methods that investigated possible publication bias only found significant result for severe left ventricular hypertrophy. However, the clinical relevance of this risk marker has already been documented in several studies and emphasized in the last guidelines.[1],[2],[11],[18],43

The limitations of our study include: (1) the inclusion and exclusion criteria, diagnostic methods, and definitions varied discreetly among the different studies; (2) the data of the patients from the same institution may have overlapped, although this did not occur in most of the analyses; (3) the absence of randomized trials may also be considered a relative limitation, but it is important to remember that the systematic review with meta-analysis has the capacity of minimizing this problem, bringing information from observational studies to a higher level of evidence; (4) myocardial fibrosis was analyzed in a few studies and in a binary manner, not quantitatively, although the latter has been gaining attention in recent publications.[14] Although the study of Chan et al.,[14] which was used in this meta-analysis, provided quantitative information, we used only the RR for the presence or absence of fibrosis.[14] Another relevant issue for discussion, although not addressed in this study, which can be considered a limitation, is the cost of cardiac MRI. No cost-effectiveness analysis aimed at the investigation of fibrosis using MRI in patients with HCM has been conducted yet.

Despite these limitations, we rely on the findings of our study because of its methodology, consistency of results (absence of heterogeneity in most analyses), and especially the close and well-known association between myocardial fibrosis and arrhythmias. And with the purpose of studying this association, it is important to emphasize that we chose to include only studies that evaluated outcomes equivalent to SD.

Conclusions

In summary, although it is very difficult to make clinical decisions of great relevance to patients based solely on information from observational studies, it is important to weigh the risks and benefits of ICDs with patients and their families. Nevertheless, this meta-analysis of observational studies emphasizes the importance of cardiac MRI in the detection of myocardial fibrosis for the risk stratification of SD in HCM and confirms the role of traditional RMs, with a doubtful role for LVOTO. Thus, new clinical prediction models using myocardial fibrosis should be considered as a primary prevention strategy for SD in these patients in the future.