The 94 total SCD events comprised 4.5% of the overall study cohort (N = 2094). There were 82 SCD events terminated among 527 patients with primary prevention implantable cardioverter/defibrillators (ICDs), 5 SCDs among 1567 patients without ICDs, and 7 resuscitated cardiac arrests among 1567 patients without ICDs. Of the 5 SCDs, 2 patients declined ICDs, 2 had no risk factors, and 1 died of left ventricular apical aneurysm before this was a recognized risk factor for SCD in hypertrophic cardiomyopathy; of the 7 patients with resuscitated cardiac arrests, 5 declined ICDs and 2 had no risk factors.
aP < .001.
A, Comparison of true sudden cardiac deaths (SCDs) using enhanced American College of Cardiology/American Heart Association (ACC/AHA) guidelines vs potential SCD events applying the European Society of Cardiology (ESC) risk model. The ESC intermediate/high–risk model assumes that patients with intermediate and high risk scores (≥4% SCD event rate over 5 years) were implanted with primary prevention implantable cardioverter/defibrillators (ICDs); the ESC high-risk model assumes that only patients with high risk scores (≥6% SCD event rate over 5 years) were implanted with ICDs. B, All-cause mortality in patients with hypertrophic cardiomyopathy (HCM) treated with primary prevention ICDs did not differ from an age-matched and sex-matched US population.
Abbreviations: ACC/AHA, American College of Cardiology/American Heart Association; ACEi, angiotensin-converting enzyme inhibitors; ARB, angiotensin receptor blocker; CAD, coronary artery disease; CMR, cardiovascular magnetic resonance imaging; EF, ejection fraction; LGE, late gadolinium enhancement; LVED, left ventricular end-diastolic dimension; LV, left ventricular; NSVT, nonsustained ventricular tachycardia; NYHA, New York Heart Association.
a Includes 21 patients with unsuccessful alcohol septal ablation prior to myectomy.
b One of 7 experienced an appropriate secondary prevention ICD intervention.
c Includes 2 patients with device infection.
d Includes 11 patients with lead fracture and inappropriate shocks.
e Most commonly cancer (n = 15) and multiple noncardiac comorbidities associated with advanced age (n = 40).
f Death related to CAD in 5 patients, postoperative aortic valve replacement in 2, and postoperative mitral valve replacement/coronary artery bypass graft in 1.
Abbreviations: +, yes; ACC/AHA, American College of Cardiology/American Heart Association; EF, ejection fraction; ESC, European Society of Cardiology; FH, family history; HF, heart failure; LA, left atrium; LV, left ventricle; LVH, left ventricular hypertrophy; LVOTG, left ventricular outflow tract gradient; MM, myectomy performed; NA, not available; NYHA, New York Heart Association; RF, risk factor; SCD, sudden cardiac death; VT, ventricular tachycardia.
a ESC risk score not calculated because patient was younger than 16 years at time of initial ICD event.
b Attenuated or hypotensive blood pressure response represents one of the risk markers.
c Myectomy after initial ICD shock.
d Myectomy prior to initial ICD shock.
e End-stage HF and consideration for heart transplant represents an indication for primary prevention ICD.
f Extensive late gadolinium enhancement on cardiovascular magnetic resonance imaging represents 1 risk marker (14.2% of LV mass in patient 40, 16.0% in patient 47, 15.2% in patient 76, and 15.2% in patient 93).
g Systolic dysfunction secondary to large apical aneurysm formation.
h Alcohol septal ablation 12 years prior to myectomy.
Abbreviation: ICD, implantable cardioverter/defibrillator.
a 51 patients younger than 16 years at initial evaluation are excluded from calculations of ESC risk scores.
b Assumes an ICD is implanted in all patients with intermediate or high ESC risk scores (≥4% event rate over 5 years).
c Assumes an ICD is implanted only in patients with high ESC risk scores (≥6% event rate over 5 years).
d As an intention to treat, which is calculated as if 2 high-risk patients who died suddenly after declining an ICD recommendation are included as theoretically receiving an appropriate ICD intervention. When these 2 patients are included as deaths, sensitivity is 87% (95% CI, 79-93).
eTable. Sudden death risk factors leading to primary prevention ICD implants in 82 patients with HCM with appropriate device interventions.
eFigure. Flow diagram summarizing clinical outcome in patients with or without primary prevention ICDs.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Maron MS, Rowin EJ, Wessler BS, et al. Enhanced American College of Cardiology/American Heart Association Strategy for Prevention of Sudden Cardiac Death in High-Risk Patients With Hypertrophic Cardiomyopathy. JAMA Cardiol. Published online May 22, 20194(7):644–657. doi:10.1001/jamacardio.2019.1391
Is it possible to identify most patients with hypertrophic cardiomyopathy (HCM) at risk of arrhythmic sudden cardiac death (SCD) and to prevent such events with prophylactic implantable cardioverter/defibrillators (ICDs)?
In this cohort study of 2094 patients with HCM, ICD decision making was prospectively assessed based on conventional major risk markers derived from the literature and enhanced from 2011 American College of Cardiology/American Heart Association guidelines over a 17-year experience at a single HCM center. Rates of appropriate ICD therapy terminating potentially lethal ventricular tachyarrhythmias exceeded SCDs in patients without ICDs by almost 50-fold.
A prospective individual risk factor strategy predicted SCD events in nearly all at-risk patients with HCM, resulting in prophylactically implanted ICDs that prevented arrhythmic SCD events.
Strategies for reliable selection of high-risk patients with hypertrophic cardiomyopathy (HCM) for prevention of sudden cardiac death (SCD) with implantable cardioverter/defibrillators (ICDs) are incompletely resolved.
To assess the reliability of SCD prediction methods leading to prophylactic ICD recommendations to reduce the number of SCDs occurring in patients with HCM.
Design, Setting, and Participants
In this observational longitudinal study, 2094 predominantly adult patients with HCM consecutively evaluated over 17 years in a large HCM clinical center were studied. All patients underwent prospective ICD decision making relying on individual major risk markers derived from the HCM literature and an enhanced American College of Cardiology/American Heart Association (ACC/AHA) guidelines–based risk factor algorithm with complete clinical outcome follow-up. Data were collected from June 2017 to February 2018, and data were analyzed from February to July 2018.
Main Outcomes and Measures
Arrhythmic SCD or appropriate ICD intervention for ventricular tachycardia or ventricular fibrillation.
Of the 2094 study patients, 1313 (62.7%) were male, and the mean (SD) age was 51 (17) years. Of 527 patients with primary prevention ICDs implanted based on 1 or more major risk markers, 82 (15.6%) experienced device therapy–terminated ventricular tachycardia or ventricular fibrillation episodes, which exceeded the 5 HCM-related SCDs occurring among 1567 patients without ICDs (0.3%), including 2 who declined device therapy, by 49-fold (95% CI, 20-119; P = .001). Cumulative 5-year probability of an appropriate ICD intervention was 10.5% (95% CI, 8.0-13.5). The enhanced ACC/AHA clinical risk factor strategy was highly sensitive for predicting SCD events (range, 87%-95%) but less specific for identifying patients without SCD events (78%). The C statistic calculated for enhanced ACC/AHA guidelines was 0.81 (95% CI, 0.77-0.85), demonstrating good discrimination between patients who did or did not experience an SCD event. Compared with enhanced ACC/AHA risk factors, the European Society of Cardiology risk score retrospectively applied to the study patients was much less sensitive than the ACC/AHA criteria (34% [95% CI, 22-44] vs 95% [95% CI, 89-99]), consistent with recognizing fewer high-risk patients.
Conclusions and Relevance
A systematic enhanced ACC/AHA guideline and practice-based risk factor strategy prospectively predicted SCD events in nearly all at-risk patients with HCM, resulting in prophylactically implanted ICDs that prevented many catastrophic arrhythmic events in this at-risk population.
Over the almost 60-year history of hypertrophic cardiomyopathy (HCM), sudden cardiac death (SCD) has been the most visible and devastating disease complication, leading to recognition of this disease as a common cause of SCD in the young.1-5 However, with the introduction of implantable cardioverter/defibrillators (ICDs) to HCM almost 20 years ago,6 which have the capability of reliably terminating potentially lethal ventricular tachyarrhythmias despite greatly increased left ventricular (LV) mass and outflow obstruction, prevention of SCD has become a reality for many patients.
Risk stratification and selection of patients with HCM for prophylactic ICD therapy has evolved, with differing prediction strategies emerging.1-11 However, these developments have left many in the practicing cardiovascular community with a measure of uncertainty regarding the most reliable option to guide identification of the highest-risk patients with HCM for whom primary prevention of SCD events with ICDs is indicated.9,11,12 Therefore, to provide additional insights into this debate regarding the most efficacious strategy to identify patients with HCM for contemporary ICD therapy and SCD prevention, we tested our 17-year experience in a large institute study cohort relying largely on risk factor algorithms from an enhanced version of the 2011 American College of Cardiology/American Heart Association (ACC/AHA) guidelines and other consensus documents.4
The database of the Hypertrophic Cardiomyopathy Center and Research Institute at Tufts Medical Center, Boston, Massachusetts, was interrogated, and 2094 patients with a clinical HCM diagnosis evaluated consecutively from June 2001 (the beginning of the program) to December 2016 were identified. Data on recent comprehensive vital clinical and survival status were obtained up to December 30, 2017, by hospital visit or telephone contact with patients, family members, and referring physicians (1897 [90.6%]) or, in the absence of the most current follow-up, was confirmed by the Social Security Death Index (197 [9.4%]). For the overall study group, the mean (SD) follow-up duration from initial clinical evaluation at Tufts Medical Center to most recent contact (or death) was 4.7 (3.5) years (9107 person-years).
Clinical diagnosis of HCM was based on the 2-dimensional echocardiographic or cardiovascular magnetic resonance imaging (CMR) identification of otherwise unexplained hypertrophied nondilated LV (wall thickness ≥13 mm).1,2,4,5 Patients had been referred for targeted subspecialty evaluations, including diagnosis, risk stratification, and treatment. Patients with phenocopies of HCM (eg, Fabry disease, LAMP2 cardiomyopathy, PRKAG2, or amyloidosis) were excluded. This study was approved by the institutional review board at Tufts Medical Center, allowing retrospective review of medical records and granting a waiver of informed consent in accordance with 45 CFR 46.116(d).
Transthoracic echocardiographic studies were performed in a standard fashion.4 Continuous-wave Doppler estimated peak instantaneous outflow gradient. Cardiovascular magnetic resonance imaging studies were performed in 1128 patients with a 1.5-T clinical scanner (Philips).13 Areas of late gadolinium enhancement (LGE) were assessed, as previously described.13
Single-chamber or dual-chamber ICDs capable of antitachycardia and antibradycardia pacing were implanted in 527 patients for primary prevention based on conventional risk markers1,4 and in 7 patients for secondary prevention after resuscitated cardiac arrest. Of 527 patients receiving ICD, 483 (91.7%) received transvenous devices and 44 (8.3%) subcutaneous ICDs.
Expert electrophysiologists interrogated stored intracardiac electrogram data to adjudicate arrhythmias responsible for ICD discharges.14 Defibrillator interventions were considered appropriate when triggered by ventricular fibrillation (VF) or sustained rapid monomorphic ventricular tachycardia (VT; rate, >180/min).6,14-25 Rate cutoffs were programmed by managing electrophysiologists. Inappropriate interventions were triggered by heart rates exceeding programmed thresholds.
Major clinical markers used to judge increased risk and guide ICD recommendations were based on consensus standards and risk factors prevailing at various times during the study, ie, 2011 ACC/AHA guidelines,4 2003 ACC/European Society of Cardiology (ESC) expert consensus,26 and 2017 AHA/ACC/Heart Rhythm Society guidelines,27 as well as markers emerging more recently with the introduction of CMR to HCM practice, ie, extensive LGE13 and LV apical aneurysms.28 The risk stratification protocol presented here represents enhancement of 2011 ACC/AHA guidelines and accounts for HCM risk stratification evolution. For each study patient, risk assessment and ICD decisions were made prospectively in a standard fashion at initial visit or annual surveillance evaluations by senior cardiologists (M.S.M., E.J.R., and B.J.M.).
One or more of the following major risk markers were considered sufficient evidence of increased SCD risk within the context of the overall clinical profile of a given patient with HCM to justify recommendations for prophylactic ICDs to prevent SCD14-27:
Family history of SCD judged to be definitively or likely caused by HCM in 1 or more first-degree or other close relatives 50 years or younger.29
Left ventricular hypertrophy, specifically with wall thickness of 30 mm or greater in any segment within the chamber detected by echocardiography or CMR30; consideration for this morphologic marker was also given to borderline thicknesses of 28 mm or 29 mm in individual patients at the discretion of the treating cardiologist.
Unexplained syncope, with 1 or more otherwise unexplained recent episodes involving loss of consciousness, judged by history unlikely to be of neurocardiogenic (vasovagal) etiology, generally occurring within 5 years after evaluation.31
Nonsustained VT, defined as 3 or more repetitive brief episodes each consisting of 3 or more consecutive ventricular beats and/or 1 or more prolonged episodes (10 beats or more) at 130 beats/min or more, usually over 24 to 48 hours of ambulatory electrocardiography monitoring32,33; has greater weight when associated with another risk marker, particularly LGE.
Late gadolinium enhancement–identified fibrosis in patients with contrast CMR with diffuse and extensive distribution, either quantified (usually comprising about 15% or more of LV mass) or estimated by visual inspection to be extensive and diffuse,13 either alone or associated with other markers.
End-stage phase, defined as systolic dysfunction with ejection fraction less than 50% by echocardiography or CMR, usually in severely symptomatic patients who are potential heart transplant candidates.34
Left ventricular apical aneurysm identified by echocardiography or CMR, independent of size, with discrete thin-walled dyskinetic or akinetic segments of distal LV chamber and associated contiguous regional scarring.28
Single (or double) sarcomere protein mutations were not regarded as SCD markers, given insufficient evidence supporting a predictive association of individual variants with SCD events.1,2,4,25 In addition, after 2011, based on recognition that age older than 60 years is itself associated with low likelihood of SCD, primary prevention ICD decisions in this age group were made on a case-by-case basis only when risk markers were perceived to carry particular weight in the individual patient.35,36 Three other disease variables were used selectively to support ICD decisions when associated with 1 or more major risk markers: abnormal blood pressure response to exercise,37,38 LV outflow obstruction with gradient of 50 mm Hg or greater at rest,39 and moderate degree of LGE.13
The ESC mathematically derived quantitative risk score was calculated in accordance with O’Mahony et al40 to predict SCD event rates over 5 years.2 Based on this score, patients are stratified into 3 risk subsets for ICD recommendations: low risk (<4% over 5 years; ICD not considered); intermediate risk (4%-6% over 5 years); and high risk (≥6% over 5 years; ICD should be considered).
We performed an external validation of ESC scores focused on management decisions and outcomes for individual patients. We calculated ESC risk scores retrospectively for only study patients 16 years and older based on data obtained at initial evaluation. In cases in which initial ICD interventions occurred more than 5 years after study entry, a second score was calculated at the clinical visit just prior to the delayed ICD therapy to account for possible changes in risk profile over time.
For survival and event analyses, the fraction of patients at each follow-up interval was estimated by the Kaplan-Meier method. Expected fractions surviving after initial visit were computed by assigning probability of survival (appropriate to age and sex) based on the US general population.41 Actual and expected survival fractions were compared using 1-sample log-rank tests, which also provided estimated standardized mortality ratios and 95% CIs.
For the purpose of sensitivity and specificity analyses, we defined true-positives as individuals who received recommendations for ICDs and subsequently experienced SCD events; false-positives as patients who received recommendations for ICDs but did not subsequently experience an SCD event; false-negatives as patients who did not receive recommendations for ICDs but subsequently experienced SCD events; and true-negatives as patients who did not receive recommendations for ICDs or subsequently experience an SCD event.
Using logistic regression, we analyzed the ability of these risk stratification approaches to discriminate between patients who experienced an SCD event during follow-up and those who did not. Receiver operating characteristics curves were constructed by plotting sensitivity against 1 − specificity with the area under the curve expressed as the C statistic. The C statistic was calculated for the study cohort using either the enhanced ACC/AHA guidelines approach (ie, the presence vs absence of the established major risk factors) or the ESC risk score.
To assess any bias resulting from missing data, clinical profiles of patients with and without missing data were compared,40 and logistic regression was used to identify predictors of missingness. Data were missing for only 1 variable (nonsustained VT) in 414 patients (19.8%). Data were assumed to be missing at random and were imputed into a risk model that included clinical outcome and predictors of missingness. Consistent with Rubin’s Rules, 25 imputed data sets were created, and values were combined into a single observation for each patient.
Data are expressed as means with standard deviations for continuous variables and as numbers with percentages for categorical variables. Continuous variables were calculated using t tests or Wilcoxon rank sum tests, and categorical variables were calculated using χ2 or Fisher exact tests. All tests were 2-sided; P values less than .05 were considered significant. Study computations were performed using R version 3.4.2 (The R Foundation) and the val.prob function.
Baseline characteristics of the 2094 included patients are shown in Table 1. The mean (SD) age of patients was 51 (17) years at initial evaluation and 56 (17) years at most recent evaluation (or death); 1313 (62.7%) were male. At initial evaluation, most patients (1433 [68.4%]) were asymptomatic or mildly symptomatic (New York Heart Association functional classes I and II); 1755 of 1871 surviving patients (93.8%) were in New York Heart Association functional class I or II at most recent evaluation.
Of the 527 patients implanted prophylactically for primary prevention, 82 (15.6%) experienced 1 or more appropriate ICD interventions for VT or VF to restore sinus rhythm (Figure 1; Table 2) (eFigure in the Supplement). One-year and 5-year cumulative probability for each patient’s initial appropriate device interventions was 3.2% (95% CI, 2.1-5.4) and 10.5% (95% CI, 8.0-13.5), respectively.
Implantable cardioverter/defibrillator events terminating VT or VF (n = 82) significantly exceeded the 5 HCM-related SCDs among 1567 patients without ICDs (0.3%) by 49-fold (95% CI, 20-119; P < .001) (Figure 1). All-cause survival for the 527 patients implanted with primary prevention ICDs did not differ from an age-matched and sex-matched general US population41 (Figure 2).
Of the 82 patients treated with ICD therapy, 24 (29%) had multiple discharges (range, 2-8); 4 others had VT storms (all in patients with LV apical aneurysms). Only 1 of the ICD defibrillation shocks was in a patient with a subcutaneous device. Time from implant to first appropriate device intervention ranged from 0.2 to 16 years (mean [SD] time, 5.2 [4.6] years). Thirty-seven patients with ICD therapy (45%) had substantially delayed intervals greater than 5 years (≥10 years in 16), ranging from 14 to 16 years in 6 patients.
At first ICD therapy, the mean (SD) age of patients was 45 (16) years, including 19 who were 30 years or younger. Left ventricular outflow obstruction (gradient ≥30 mm Hg) was present at rest in 23 patients (28%). The mean (SD) maximal LV thickness was 19 (4) mm (range, 13-48 mm).
Of the 82 patients with terminated VT or VF events, 78 (95%) were alive at a mean (SD) age of 49 (16) years at a mean (SD) 4.6 (4.3) years after initial device therapy; 71 (91%) were asymptomatic or only mildly symptomatic. At the time of submission, 4 patients had died, including 3 of end-stage heart failure awaiting transplant34 and 1 of lung cancer (eFigure in the Supplement).
Of the 82 patients with appropriate ICD interventions, 53 (65%) had been implanted based on a single major risk marker, with the most common being unexplained syncope, family history of HCM-related SCD, multiple and/or prolonged nonsustained VT, massive LV hypertrophy, and LV apical aneurysm. In 20 other patients (24%), ICD decisions were based on a combination of 2 risk markers. The most common combinations of markers were unexplained syncope associated with either a family history of SCD, nonsustained VT, or massive LVH. Nine patients (11%) had 3 or 4 risk factors (eTable in the Supplement). Of patients with 1, 2, or 3 or more risk factors, the 3 recently emerging clinical markers were associated with appropriate ICD interventions in 21 patients (26%), including LV apical aneurysms (n = 12), end-stage heart failure (n = 6), and extensive LGE (n = 3).13
Five of 1567 patients without ICDs (0.3%) aged 40 to 59 years died suddenly in clinical circumstances for which HCM was the probable cause (0.1% per year) (Figure 1; Table 2). Each event occurred under sedentary conditions or with modest physical exertion, and 4 of the 5 deaths occurred before 2010.
Of the 5 patients with SCD, 2 had 2 risk factors each but declined the recommendation of a prophylactic ICD (Figure 1). Two other patients, aged 45 and 55 years, were without conventional risk factors, although neither underwent CMR imaging or postmortem examination.13 The remaining patient was identified with an LV apical aneurysm (in 2004) well before this abnormality became an established risk marker and an ICD indication in HCM.42 Other deaths in the cohort, predominantly noncardiac in nature, are summarized in Table 1.
Seven of 1567 patients without ICDs (0.4%) survived out-of-hospital resuscitated cardiac arrest over a mean (SD) time of 2.8 (1.8) years after initial evaluation (0.1% per year) (Table 2). Five of these patients had 1 or 2 conventional risk factors but declined recommendations for prophylactic ICDs; 2 others had no risk markers to justify an ICD. All 7 patients received secondary prevention ICDs, and 1 has experienced an appropriate device intervention.
Of the 527 patients with primary prevention ICDs, 85 (16.1%) experienced device-related complications, at a rate of 2.1% per year. Predominantly, these were inappropriate shocks (n = 61), lead fractures (n = 21), or infection (n = 14). The complication rate in patients with appropriate device therapy did not differ from that of patients with ICDs but without device interventions (1.9% vs 2.2%).
The C statistic calculated for the study cohort using enhanced 2011 ACC/AHA guidelines was 0.81 (95% CI, 0.77-0.85), discriminating between patients with and without SCD events. Sensitivity for predicting SCD events was 87% (95% CI, 79-93) and as an intention to treat was 95% (95% CI, 89-99) (Table 3). Specificity for predicting patients without events was 78% (95% CI, 76-80). The number of patients needed to treat with ICDs to appropriately treat 1 patient with terminated VT or VF was 6.6 (95% CI, 5.5-8.2).
The C statistic calculated for the study cohort using the ESC model was 0.74 (95% CI, 0.69-0.80). The sensitivity for predicting SCD events with high-risk scores of 6% or greater over 5 years was 34% (95% CI, 24-44), but the specificity for predicting patients without events was 92% (95% CI, 91-94) (Table 3).
Combined ESC high-risk and intermediate-risk scores of 4% or greater over 5 years had a sensitivity of 58% (95% CI, 47-69) and a specificity of 81% (95% CI, 79-82). The number of patients needed to treat with ICDs to appropriately treat 1 patient with terminated VT or VF was 10.3 (95% CI, 7.8-15.3).
Among the 91 patients with SCD events (ie, appropriate primary prevention ICD interventions, SCDs, and resuscitated cardiac arrests at 16 years and older), only 31 (34%) had high-risk ESC scores (≥6% over 5 years) considered sufficient to recommend a prophylactic ICD (Table 2 and Table 3). Also, 38 of 91 patients with SCD events (42%) had low-risk scores that likely would have excluded an ICD recommendation (Table 2).
The highly visible specter of unpredictable SCD in young people due to HCM has dominated perceptions of this heterogeneous genetic disease in patient and physician communities for decades.1-5,25 Increasing penetration of the ICD into the HCM population over the last 18 years6 has prevented SCD events and also provided extended longevity with good quality of life and psychological well-being17 for many patients, largely free of later morbidity and mortality.1,3,4,6,14-25 Indeed, the efficacy of the ICD has now been reported in several hundred patients with HCM representing the broad clinical spectrum of the disease,14-25 largely in the United States and Europe, selected predominantly with reliance on the conventional risk factors included in prior consensus guidelines.2,4,26,27 However, given the recent evolution of HCM risk stratification, including the emergence of novel risk markers and also a score-based prediction model,2,40,43,44 the optimal strategy for identifying high-risk patients who will benefit from prophylactic ICD therapy deserves further scrutiny.45
Therefore, to assess the reliability of SCD prediction strategies, we assembled the present study group from our large, long-standing multidisciplinary HCM center with prospective decision making focused on the outcome and therapy of individual patients. Over a 17-year period, clinical practice patterns and strategies used in our center resulted in standardized decisions for primary prevention ICDs by relying on the presence of 1 or more acknowledged major risk factors within the context of the overall clinical profile of the patient, as derived from the HCM literature2,4,26,27 as well as clinical markers that have emerged over the last 5 years.1,13,28,42 All devices were implanted in accord with the overall clinical profile and wishes of the individual patient during the ICD era for HCM1,3,6,14,17 and using risk stratification markers easily translatable to cardiovascular practice outside of specialized HCM center environments.
We are able to substantiate the potential power of the ICD in changing the clinical course of HCM1,3 by reporting a highly positive outcome in which potentially lethal ventricular tachyarrhythmias were terminated by ICDs in 16% of our cohort implanted for primary prevention. Notably, this result is in sharp contrast (by almost 50-fold) to the very low frequency of HCM-related SCDs occurring in patients in our cohort not implanted with ICDs. Of the patients not implanted with ICDs who died, 40% were identified with SCD markers but nevertheless declined ICD recommendations. Furthermore, just 2 of the 2094 study patients (0.2%) experienced SCD events without risk markers, a rate similar to that expected in the general population and consistent with prior recognition that a small minority of patients with HCM experience arrhythmic events without evidence suggesting susceptibility to SCD.46
As part of this analysis, we also calculated a C statistic for enhanced ACC/AHA risk markers that demonstrated reasonably good discrimination between patients who would or would not experience SCD events (high risk vs low risk). However, since certain aspects of risk stratification performance may not be reflected in the C statistic alone,47 we elected to emphasize the prospective prediction of individual patient outcomes driven by prophylactic ICD implant decisions.48
In this regard, the ACC/AHA strategy proved to be superior to the mathematical ESC risk score and highly sensitive for predicting events terminating potentially lethal ventricular tachyarrhythmias,49 ie, 87% (95% CI, 79-93) or 95% (95% CI, 89-99) as intention to treat. Therefore, we were able to prospectively identify about 95% of those patients who developed potentially lethal ventricular tachyarrhythmias.
The ESC model is associated with higher specificity for identifying patients unlikely to have events, potentially decreasing implants in low-risk patients. However, sensitivity of the ESC score for identifying individual patients at risk of SCD events was only 34%, ie, only one-third of our patients implanted based on individual ACC/AHA markers and with appropriate device therapy terminating ventricular tachyarrhythmias would have received high ESC risk scores (≥6% over 5 years) sufficient to justify primary prevention ICDs. The remaining patients would have been left unprotected from SCD without ICDs. Therefore, enhanced ACC/AHA guidelines are consistent with the dominant principle of risk stratification in HCM, which is the prevention of SCD.49
We should underscore that flexibility inherent in the ACC/AHA strategy (unencumbered by rigid automated scoring) allowed us the liberty to selectively modify risk stratification as our clinical perceptions evolved over long periods of time. In particular, these changes permitted incorporation of novel high-risk markers, such as LV apical aneurysms28,36,42 and extensive LGE13 (both related to penetration of CMR into the HCM population and both absent from the ESC risk score) as well as systolic dysfunction. Together, these risk markers were associated with more than 25% of appropriate ICD therapies in this study (although notably, none of these markers are part of the ESC risk score formula).2,40 Also, consistent with best medical practice,50 our initiative permitted a measure of clinical reasoning, individual physician judgment, and a patient-preference shared decision-making narrative,51,52 important for identifying those patients requiring an ICD with a complex, heterogeneous disease, such as HCM.1,3-5,25
Not unexpectedly, benefits attributable to ICDs for patients with HCM1,3,6,17 come at some cost, such as potential for device complications and inconvenience, a consideration now mitigated by the relatively low device-related complication rate reported here (15%), which is identical to the proportion of patients with appropriate ICD interventions and is also likely to decrease further given advancements in device programming. A related concern is the possibility of overtreatment with ICDs that do not intervene. Nevertheless, the number of patients needed to treat with ICDs to appropriately treat 1 patient with VT or VF was similar using both enhanced ACC/AHA and ESC guidelines and was not substantially different from randomized clinical trials of ICDs in ischemic and nonischemic cardiomyopathy53-55 or nonrandomized HCM studies.6,16,40,43,44
Notably, apparent ICD excess does not always represent unnecessary implants, since it is not uncommon for prophylactic ICDs to unpredictably terminate life-threatening ventricular tachyarrhythmias over extended periods.3,14,17,56 Indeed, in the present study population, a large proportion of ICD therapies occurred more than 5 years after implant (up to 16 years), and we would estimate that about 20% of devices not yet activated could do so over the next 10 years.
There are potential study limitations that justify mention here. First, there can be some concern that rapid monomorphic VT episodes terminated by ICD shocks may not be truly life threatening. This possibility originally arose from observations in randomized clinical trials of ICDs in non-HCM diseases for which there was an excess of ICD shocks compared with true SCD events, thereby providing the theoretical possibility that rapid VT could have terminated spontaneously were the ICD not present.57 However, because there is no evidence resolving this issue specifically in HCM,3,6,14,19-21 we prefer not to make assumptions in this regard.
Second, that the present large study cohort has a single-center design is not necessarily a limitation but rather advantageous in some respects. Clinical strengths of single-center studies include allowing for well-characterized and vigorous follow-up of individual patients and standardized management decisions not as easily achieved in multicenter studies, particularly for heterogeneous diseases such as HCM.
We report a highly effective clinical practice strategy for prevention of SCD in HCM based on 1 or more conventional risk markers considered major within individual patient profiles. These data reflect the 17-year experience of a multidisciplinary HCM center and demonstrate what is possible for preventing SCD and lowering mortality in the contemporary management era of this disease.58 Although there will always be some inherent uncertainty in precisely predicting future HCM-related events for all patients by any risk stratification method, nevertheless, the SCD prevention approach reported here represents a treatment paradigm affording the potential to prevent nearly all SCDs in an outpatient HCM population.
Accepted for Publication: March 11, 2019.
Corresponding Author: Martin S. Maron, MD, Hypertrophic Cardiomyopathy Center and Research Institute, Division of Cardiology, Tufts Medical Center, 800 Washington St, Boston, MA 02111 (firstname.lastname@example.org).
Published Online: May 22, 2019. doi:10.1001/jamacardio.2019.1391
Author Contributions: Drs M. Maron and B. Maron had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: M. Maron, Rowin, Fatima, Link, B. Maron.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: M. Maron, Rowin, Mooney, Fatima, B. Maron.
Critical revision of the manuscript for important intellectual content: M. Maron, Rowin, Wessler, Patel, Koethe, Romashko, Link, B. Maron.
Statistical analysis: Wessler, Fatima, Patel, Koethe.
Obtained funding: B. Maron.
Administrative, technical, or material support: M. Maron, Fatima, Patel, Romashko.
Study supervision: M. Maron, Rowin, B. Maron.
Conflict of Interest Disclosures: Dr Wessler has received grants from the American College of Cardiology and National Institutes of Health. No other disclosures were reported.
Additional Contributions: We acknowledge the generous support of Jim and Gina Penn, who donated personal funds to this project. They were not compensated for their contribution.
Create a personal account or sign in to: