-
PDF
- Split View
-
Views
-
Cite
Cite
Marcello Disertori, Silvia Quintarelli, Silvia Mazzola, Valentina Favalli, Nupoor Narula, Eloisa Arbustini, The need to modify patient selection to improve the benefits of implantable cardioverter-defibrillator for primary prevention of sudden death in non-ischaemic dilated cardiomyopathy, EP Europace, Volume 15, Issue 12, December 2013, Pages 1693–1701, https://doi.org/10.1093/europace/eut228
- Share Icon Share
Abstract
Left ventricular ejection fraction (LVEF) ≤35% is a major determinant for implantable cardioverter-defibrillator (ICD) therapy for primary prevention of sudden death (SD) in patients with non-ischaemic dilated cardiomyopathy (DCM). However, as a risk marker for SD, low LVEF has limited sensibility and specificity. Selecting patients according to the current guidelines shows that most DCM patients do not actually benefit from ICD implantation and may suffer collateral effects and that many patients who are at risk of SD are not identified because a large proportion of SD patients exhibit only mildly depressed LVEF. Identifying patients who are at risk of SD on the sole basis of LVEF appears to be an over-simplification which does not maximize the benefit of ICD therapy. Owing to the complexity of the substrates underlying SD, multiple risk factors used in combination could probably predict the risk of SD better than any individual risk marker. Among non-invasive tests, microvolt T-wave alternans and cardiac magnetic resonance with late gadolinium enhancement may contribute to a better SD risk stratification by their high negative predictive value. Genetics may further contribute because approximately one-third of DCM patients have evidence of familial disease, and mutations in some known disease genes, including LMNA, have been associated with a high risk of SD. In this review, we critically analyse the current indications for ICD implantation and we explore existing knowledge about potentially predicting markers for selecting DCM patients who are at high and low risk of SD.
Introduction
Thirty years after the introduction of the implantable cardioverter-defibrillator (ICD) by Mirowski et al.,1 this device is now widely utilized in clinical practice, and its efficacy has been proven in a number of studies.2–5 The use of the ICD has reduced the incidences of sudden death (SD) and total mortality in a variety of clinical scenarios, in both primary and secondary prevention settings.4 A debate is currently ongoing regarding the selection of patients for ICD therapy for primary prevention of SD;6–11 patients who are affected by non-ischaemic dilated cardiomyopathy (DCM) are a particularly pressing concern.11
At the moment, the Guidelines for ICD implantations in DCM patients are still based on left ventricular ejection fraction (LVEF) values and New York Heart Association (NYHA) functional classes (Table 1).12–14 There is no evidence that identifying patients who have a high risk of SD on the sole basis of EF and NYHA class allows the best exploitation of the benefits of ICD therapy.8,11 The goal is to identify the patients who will benefit the most from ICD therapy and to avoid exposing other patients to unnecessary procedural risk. Many DCM patients who die from SD had only a moderately depressed LVEF.15,16 In addition, ∼80% of the patients with DCM and LVEF ≤ 35% who received an ICD did not show therapeutic ICD intervention during a 5-year follow-up.3 Therefore, LVEF demonstrates low sensitivity and specificity17 and is, by itself, not a sufficient criterion for prognostic stratification of SD risk and ICD decision-making in patients with DCM.
Guideline . | DCM patients . | Recomm. and evidence; appr. use score (1–9) . | OMT . | Contraindications . |
---|---|---|---|---|
ACCF/AHA/HRS Guidelines 201212 | LVEF ≤ 35%, NYHA classes II–III | Class I; LoE B | OMT (duration not defined) | L. expect. < 1 year; important comorbidities |
ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR Appropriate use criteria 201314 | LVEF ≤ 35%, NYHA classes II–III | A (9) | At least 3 months on OMT | L. expect. < 1 year; important comorbidities |
LVEF ≤ 35%, NYHA class I | A (7) |
Guideline . | DCM patients . | Recomm. and evidence; appr. use score (1–9) . | OMT . | Contraindications . |
---|---|---|---|---|
ACCF/AHA/HRS Guidelines 201212 | LVEF ≤ 35%, NYHA classes II–III | Class I; LoE B | OMT (duration not defined) | L. expect. < 1 year; important comorbidities |
ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR Appropriate use criteria 201314 | LVEF ≤ 35%, NYHA classes II–III | A (9) | At least 3 months on OMT | L. expect. < 1 year; important comorbidities |
LVEF ≤ 35%, NYHA class I | A (7) |
DCM, non-ischaemic dilated cardiomyopathy; Recomm., recommendation; appr., appropriate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; LoE, level of evidence; OMT, optimal medical therapy; L. expect., life expectancy.
Guideline . | DCM patients . | Recomm. and evidence; appr. use score (1–9) . | OMT . | Contraindications . |
---|---|---|---|---|
ACCF/AHA/HRS Guidelines 201212 | LVEF ≤ 35%, NYHA classes II–III | Class I; LoE B | OMT (duration not defined) | L. expect. < 1 year; important comorbidities |
ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR Appropriate use criteria 201314 | LVEF ≤ 35%, NYHA classes II–III | A (9) | At least 3 months on OMT | L. expect. < 1 year; important comorbidities |
LVEF ≤ 35%, NYHA class I | A (7) |
Guideline . | DCM patients . | Recomm. and evidence; appr. use score (1–9) . | OMT . | Contraindications . |
---|---|---|---|---|
ACCF/AHA/HRS Guidelines 201212 | LVEF ≤ 35%, NYHA classes II–III | Class I; LoE B | OMT (duration not defined) | L. expect. < 1 year; important comorbidities |
ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR Appropriate use criteria 201314 | LVEF ≤ 35%, NYHA classes II–III | A (9) | At least 3 months on OMT | L. expect. < 1 year; important comorbidities |
LVEF ≤ 35%, NYHA class I | A (7) |
DCM, non-ischaemic dilated cardiomyopathy; Recomm., recommendation; appr., appropriate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; LoE, level of evidence; OMT, optimal medical therapy; L. expect., life expectancy.
During the last 2 years, many studies were published on the use of SD risk markers that are different from the LVEF and NYHA classes, and they have influenced this debate. In 2011 and 2012, two important meta-analyses on microvolt T-wave alternans (TWA) were published by Gupta et al.18 and by Merchant et al.,19 both confirming the very low annualized risk of SD in DCM patient with negative test (1.35 and 0.9%, respectively), particularly if the test is performed on beta-blocker treatment.20 Furthermore, very recent studies21–27 and one meta-analysis28 on cardiac magnetic resonance with late gadolinium enhancement (LGE-CMR) confirmed low arrhythmic risk in DCM patients with a negative test. In an additional recent prospective trial of 472 patients with DCM,27 the arrhythmic composite endpoint [SD, appropriate ICD discharge, and non-fatal ventricular tachycardia/fibrillation (VT/VF)] was reached in 7% (1.31% per year) of patients without midwall fibrosis in a median duration of follow-up of 5.3 years.
In this review, we revise present indications to ICD implantation and discuss existing and robust data with potential prognostic effect which could enter a multitask list of novel criteria for stratifying DCM patients who are at high and low risk of SD.
The scenario with current implantable cardioverter-defibrillator therapy guidelines
Non-ischaemic dilated cardiomyopathy classification
In the classification of cardiomyopathies, DCM is defined by the presence of left ventricular (LV) dilatation and LV systolic dysfunction in the absence of abnormal loading conditions or coronary artery disease sufficient to cause global systolic impairment. Right ventricular dilatation and dysfunction may be present but are not necessary for the diagnosis.29 Non-ischaemic dilated cardiomyopathy may be the consequence of a wide variety of causes, including virus-mediated disease, immune deregulation, toxic, metabolic, and inherited conditions. Unfortunately, LV enlargement and dysfunction represent the final common pathway of different disease entities and are rarely sufficient as criteria for a definite diagnosis. At least 25% of patients with DCM have evidence for familial disease with predominantly autosomal dominant inheritance.29 The arrhythmic mechanisms underlying DCM, and as consequence the risk of SD, may be different in different aetiologies. For example, a higher risk of life-threatening ventricular arrhythmias has been recently reported in patients with cardiac sarcoidosis.30,31 Also, the risk of SD is significantly higher in patients with DCM due to a Lamin A/C mutation32 compared with patients with DCM due to Dystrophin mutation.33 The SD risk seems to be lower in DCM patients compared with ischaemic dilated cardiomyopathy patients.3,34 Improvements of LVEF have been observed in >50% of patients with recent onset DCM after a period of 3–9 months on optimal medical therapy (OMT) (including beta-blockers).35–37 These data support delaying ICD implantation for a trial of medical therapy in the majority of DCM patients.
Benefits from implantable cardioverter-defibrillator primary prevention therapy in patients with non-ischaemic dilated cardiomyopathy
Four randomized controlled trials (RCTs) analysed ICD therapy, without cardiac resynchronization therapy (CRT), in DCM patients (Table 2). In both CAT (Cardiomyopathy Trial)38 and SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial),3 patients were enroled only on the basis of LVEF (≤30 and ≤35%, respectively) and NYHA class II or III. The CAT, which randomized 104 patients to receive an ICD or medical treatment, was stopped prematurely due to statistical futility in reaching the primary endpoint of reduced total mortality. The SCD-HeFT trial enroled 2521 ischaemic and DCM (48%) patients who were randomized to conventional therapy for heart failure (HF) plus placebo, conventional therapy plus amiodarone, or conventional therapy plus a conservatively programmed, shock-only, single-lead ICD, with a median follow-up of 45.5 months. Implantable cardioverter-defibrillator therapy reduced the overall mortality by 23% (P = 0.007) compared with placebo, with an absolute decrease in mortality of 7.2% after 5 years (1.4% per year). Subgroup analysis revealed that ICD therapy reduced the mortality rate compared with placebo in DCM patients as well, but this reduction was not significant [hazard ratio (HR): 0.73; 95% confidence interval (CI): 0.50–1.07; P = 0.06].
Study . | Inclusion criteria . | DCM pts (n) . | Treatment . | Mean FU m. . | Hazard ratio . | P-value . |
---|---|---|---|---|---|---|
Cat38 2002 | LVEF ≤ 30%, NYHA II–III | 104 | OMT vs. ICD | 66 | – | 0.55 |
AMIOVIRT39 2003 | LVEF ≤ 35%; NSVT; NYHA I–III | 103 | A vs. ICD | 24 | – | 0.80 |
DEFINITE40 2004 | LVEF ≤ 35%; NSVT; PVE; NYHA I–III | 458 | OMT vs. ICD | 29 | 95% CI 0.65 (0.40–1.06) | 0.08 |
SCD-HeFT DCM subgroup3 2005 | LVEF ≤ 35%; NYHA II–III | 1211 | OMT vs. OMT + A vs. OMT + ICD | 45.5a | 97.5% CI 0.73 (0.50–1.07) | 0.06 |
Study . | Inclusion criteria . | DCM pts (n) . | Treatment . | Mean FU m. . | Hazard ratio . | P-value . |
---|---|---|---|---|---|---|
Cat38 2002 | LVEF ≤ 30%, NYHA II–III | 104 | OMT vs. ICD | 66 | – | 0.55 |
AMIOVIRT39 2003 | LVEF ≤ 35%; NSVT; NYHA I–III | 103 | A vs. ICD | 24 | – | 0.80 |
DEFINITE40 2004 | LVEF ≤ 35%; NSVT; PVE; NYHA I–III | 458 | OMT vs. ICD | 29 | 95% CI 0.65 (0.40–1.06) | 0.08 |
SCD-HeFT DCM subgroup3 2005 | LVEF ≤ 35%; NYHA II–III | 1211 | OMT vs. OMT + A vs. OMT + ICD | 45.5a | 97.5% CI 0.73 (0.50–1.07) | 0.06 |
The primary endpoint was total mortality.
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association classes; OMT, optimal medical therapy; ICD, implantable cardioverter-defibrillator; NSVT, non-sustained ventricular tachycardia; A, amiodarone; PVE, premature ventricular ectopy; CI, confidence interval.
aMedian FU months.
Study . | Inclusion criteria . | DCM pts (n) . | Treatment . | Mean FU m. . | Hazard ratio . | P-value . |
---|---|---|---|---|---|---|
Cat38 2002 | LVEF ≤ 30%, NYHA II–III | 104 | OMT vs. ICD | 66 | – | 0.55 |
AMIOVIRT39 2003 | LVEF ≤ 35%; NSVT; NYHA I–III | 103 | A vs. ICD | 24 | – | 0.80 |
DEFINITE40 2004 | LVEF ≤ 35%; NSVT; PVE; NYHA I–III | 458 | OMT vs. ICD | 29 | 95% CI 0.65 (0.40–1.06) | 0.08 |
SCD-HeFT DCM subgroup3 2005 | LVEF ≤ 35%; NYHA II–III | 1211 | OMT vs. OMT + A vs. OMT + ICD | 45.5a | 97.5% CI 0.73 (0.50–1.07) | 0.06 |
Study . | Inclusion criteria . | DCM pts (n) . | Treatment . | Mean FU m. . | Hazard ratio . | P-value . |
---|---|---|---|---|---|---|
Cat38 2002 | LVEF ≤ 30%, NYHA II–III | 104 | OMT vs. ICD | 66 | – | 0.55 |
AMIOVIRT39 2003 | LVEF ≤ 35%; NSVT; NYHA I–III | 103 | A vs. ICD | 24 | – | 0.80 |
DEFINITE40 2004 | LVEF ≤ 35%; NSVT; PVE; NYHA I–III | 458 | OMT vs. ICD | 29 | 95% CI 0.65 (0.40–1.06) | 0.08 |
SCD-HeFT DCM subgroup3 2005 | LVEF ≤ 35%; NYHA II–III | 1211 | OMT vs. OMT + A vs. OMT + ICD | 45.5a | 97.5% CI 0.73 (0.50–1.07) | 0.06 |
The primary endpoint was total mortality.
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association classes; OMT, optimal medical therapy; ICD, implantable cardioverter-defibrillator; NSVT, non-sustained ventricular tachycardia; A, amiodarone; PVE, premature ventricular ectopy; CI, confidence interval.
aMedian FU months.
The AMIOVIRT (Amiodarone vs. Implantable Cardioverter-Defibrillator)39 and DEFINITE (Defibrillators in Non-Ischaemic Cardiomyopathy Treatment Evaluation)40 trials enroled patients with DCM, LVEF ≤ 35%, NYHA classes I–III, non-sustained ventricular tachycardia (NSVT), and frequent premature ventricular ectopy (PVE) in DEFINITE; both trials randomized the patients to ICD placement vs. amiodarone (AMIOVIRT) or OMT (DEFINITE). Like the CAT, the AMIOVIRT trial (103 patients) was stopped prematurely due to lack of statistical significance in reaching the primary endpoint of reduced total mortality. In the DEFINITE trial (458 patients), during a mean follow-up of 29 ± 14 months, despite a significant reduction of SD rate in the ICD group compared with the OMT group (P = 0.006), all-cause mortality did not significantly decrease (HR: 0.65; 95% CI: 0.40–1.06; P = 0.08).
The ongoing DANISH (Danish ICD study in patients with dilated cardiomyopathy, Clinical trial.gov identifier: NCT00542945) trial will randomize 1000 patients, with DCM and LVEF ≤ 35%, to ICD therapy vs. medical treatment, with primary endpoint of total mortality and a treatment phase of a minimum of 36 months. The trial started in 2007 but the results are not available.
Therefore, in DCM, ICD may reduce the risk of SD but total mortality is not significantly modified. We emphasize that in all these trials, patients were predominantly selected on the basis of low LVEF. Only the meta-analyses41,42 suggested a significant reduction in total mortality among patients who were randomized to ICD vs. medical treatment [relative risk (RR) 0.74; 95% CI: 0.59–0.93, P = 0.009].41
Left ventricular ejection fraction as risk stratification tool: lack of sensitivity and specificity
Mortality risk in patients with LV dysfunction is heterogeneous: evolution towards SD or HF or non-cardiovascular death from comorbidity depends on many variables. Low LVEF identifies a group of patients with a relatively increased risk of SD, but the sensitivity and specificity are low.17,43 When SD is measured as an absolute number of events per year, only a minority of events occurs in patients with severe LV dysfunction.15 In the Maastricht Circulatory Arrest Registry44 of 492 SD victims, 200 victims had echocardiographic data on LV function. The risk of SD did not significantly differ in subjects with severely depressed LVEF compared with those subjects with moderately depressed LVEF. Moreover, in the community-wide study of Stecker et al.,16 only one-third of SD cases had LV dysfunction which met the criteria for prophylactic ICD implantation.
Finally, among patients selected for ICD therapy on the basis of LVEF cut-off (30–35%), two-thirds to three-quarters of ICD recipients in the observational studies received no therapeutic ICD discharges, and only 5–7% of RCT participants with DCM received an appropriate shock per year.3,4
Adverse effects of implantable cardioverter-defibrillator
Implantable cardioverter-defibrillator implantation presents an iatrogenic risk that increases with the complexity of the device. In a meta-analysis by Adabag et al.45 that evaluated 4317 patients who were implanted with an ICD or ICD/CRT, there was 4% incidence of adverse events at 30 days post-implantation in the ICD group vs. 18% in the ICD/CRT group, mostly due to increased numbers of lead dislodgement, implant failure, or infection. Moreover, in the large REPLACE registry,46 there were 4–15% major complications at 6 months in ICD replacements or upgrade procedures to CRT. These complications are front-loaded adverse effects, while benefit often takes several years to appear. Implantable cardioverter-defibrillator-related complications may appear during follow-up as well: there is a high percentage of inappropriate shocks (10–20%), which impart a negative effect on both quality of life and mortality.4,47 Similarly, a perfectly working ICD is not always able to stop ventricular arrhythmias and can sometimes even be pro-arrhythmic.48 In the SCD-HeFT, 20% of deaths in the ICD group were deemed to be SD, probably due to arrhythmias.49
The recall advisories for suspected ICD or catheter dysfunction led to a decrease in quality of life. Almost all ICD producers have been involved in class I recall advisories (a designation indicating a risk of serious injury or death) in the last 10 years, requiring short-term ICD controls in a large number of patients. However, the implementation of telemetric monitoring systems and algorithms to prevent unnecessary therapies would be expected to lower ICD revisions relating to recall advisories.
Finally, as a paradox, ICD may shift SD risk in HF death risk. In patients with advanced HF, recurrent VT/VF can be a sign of progression of the impaired LV function. Although the ICD can successfully treat VT/VF, it cannot prevent death from pump failure.
New markers to stratify the risk of sudden death
Sudden death in familial dilated cardiomyopathies
At least 25% of DCM cases are of familial origin.29 Knowledge progression in the genetic basis of DCM is increasing the number of patients for whom the causative mutation is identified. In 2011, Hershberger and Siegfried50 reported that 33 genes, 31 autosomal and 2 X-linked, are causally associated with DCM. In addition, one specific mutation can be responsible for different phenotypes, while one specific phenotype may be associated with mutations in different genes. Large studies documenting the correlation between gene mutations and SD risk are lacking. Therefore, for prognostic stratification of patients who are candidates for ICD implantation, it is useful to at least consider the more common disease genes and mutations which are associated with a well-known phenotype.
Mutations in the LMNA gene, which encodes Lamin A/C, are involved in 8% of familial DCM and in 2% of sporadic DCM.50 Pasotti et al.51 reported that DCM patients with LMNA defects are at risk for HF and SD due to life-threatening arrhythmias. In a recent multicentre study,32 a cohort of 269 LMNA mutation carriers was evaluated for risk factors for malignant ventricular arrhythmias (MVAs). After a median follow-up of 43 months, independent risk factors for MVA were NSVT, LVEF < 45% at the first clinical appointment, male sex, and non-missense mutations. Malignant ventricular arrhythmia occurred only in patients with at least two of these risk factors. Recent data show that the cardiac phenotypes associated with LMNA defects may mimic arrhythmogenic right ventricular cardiomyopathy (ARVC), thus highlighting how arrhythmogenic risk is non-obligatorily related to LV dysfunction.52 On the contrary, male patients with X-linked DCM related to mutations in Dystrophin had a high risk of end-stage HF compared with a lower risk of MVA.33 Mutations in genes that typically cause ARVC may also cause dilated cardiomyopathy,53 with typical evolution through end-stage HF. The emerging spectrum of genotype–phenotype correlations in inherited DCM is far more complex than expected; thus, in this novel scenario, indications to ICD implantation should be reconsidered for certain subgroups of patients.
In the Heart Failure Society of American Practice Guidelines, Hershberger et al.54 suggested considering LMNA genetic testing for all DCM patients. This suggestion is reasonable, and the test is feasible in most molecular genetic labs, after appropriate genetic counselling. After a positive genetic test, the patient's family members should be screened for presence of the same mutation. Therefore, we encourage including genetic data in the decision-making algorithm for ICD therapy as a primary prevention at least in patients with familial DCM, as shown in Figure 1. However, this suggestion should be validated in a large study.
![Sudden death (SD) primary prevention in DCM patients: a decision-making algorithm. OMT, optimal medical therapy; OMT may vary from patient to patient but should include both beta-blockers and angiotensin-converting enzyme inhibitors or angiotensin II-receptor blockers, at the highest tolerated dosage, and should last enough to improve and stabilize clinical status.14 Important comorbidities are age, diabetes, renal failure, advanced form of chronic pulmonary failure, dementia, neoplasia, cirrhosis, or stroke (in this population, short- and mid-term non-arrhythmic mortality could largely prevail, abolishing the ICD benefit). LVEF, left ventricular ejection fraction. TWA abnormal, microvolt T-wave alternans test, either positive or indeterminate. LGE-CMR fibrosis, late gadolinium enhancement-cardiac magnetic resonance midwall fibrosis.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/europace/15/12/10.1093_europace_eut228/3/m_eut22801.jpeg?Expires=1722538975&Signature=d0Ynd7aVqZalnqAYeVKgjT5bwFyq34JcHpoLsFUbpVqoQH~0JE62UPnTzuH4biu6Ags8vZrN5nUqJSqmSizSNtEhMpEZLfuIcqeAFlzScmMGy0imAx-4A5Y5YwYMZoekiyqAl-MKFP38KUGB30r0~SOxaDSKqCy10ZDx9unocCrKqewxkU0L84qgefWYf3Lf44W4Kk4nh6V61oy8cRBHszX9tOuM5j1o-q4AnnJSqO3365kZd3x~~Iz5PpHYtbE7fueehmv0kmn~c8ir5JH-d04NU6ZbKrrk0wqWKWqx4ekk7bnRUgF7EEscX2fgjPm~Oj7q1QyBqDiRqvPehGT2EQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Sudden death (SD) primary prevention in DCM patients: a decision-making algorithm. OMT, optimal medical therapy; OMT may vary from patient to patient but should include both beta-blockers and angiotensin-converting enzyme inhibitors or angiotensin II-receptor blockers, at the highest tolerated dosage, and should last enough to improve and stabilize clinical status.14 Important comorbidities are age, diabetes, renal failure, advanced form of chronic pulmonary failure, dementia, neoplasia, cirrhosis, or stroke (in this population, short- and mid-term non-arrhythmic mortality could largely prevail, abolishing the ICD benefit). LVEF, left ventricular ejection fraction. TWA abnormal, microvolt T-wave alternans test, either positive or indeterminate. LGE-CMR fibrosis, late gadolinium enhancement-cardiac magnetic resonance midwall fibrosis.
In contrast, the 2013 report of ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR on the appropriate use criteria for implantable ICDs14 stated that patients with familial DCM which is associated with SD and evidence of structural cardiac disease but LVEF > 35% have an appropriate ICD indication, irrespective of their specific gene mutation.
May risk stratification benefit from a multiple risk markers model?
During the last decade, great efforts have been made for identifying risk markers to stratify SD risk. QRS duration, QT interval, and QT dispersion, signal-average electrocardiogram (ECG), heart rate variability, baroreflex sensitivity, heart rate turbulence, presence of PVEs or NSVT, have been studied. However, the amount of data are generally limited in DCM patients.43 None of these markers has been considered as robust as LVEF in the scientific statement of AHA/ACCF/HRS on non-invasive risk stratification techniques for identifying patients at risk for sudden cardiac death.43 Fragmented QRS on 12-lead ECG seemed a promising predictor of arrhythmic events in DCM patients,55 but the data were not confirmed in a recent study.56 In contrast to patients with ischaemic cardiomyopathy, invasive electrophysiological study with programmed ventricular stimulation did not prove to be useful for SD risk stratification in patients with DCM.57 Finally, clinical and laboratory score systems were validated to predict 1-year mortality; patients with an expected high 1-year mortality could not benefit from ICD therapy.58,59
While no single marker has been proven to be as robust as LVEF,17,43 the combination of LVEF with other markers identifies high- and low-risk patients in a better manner.6,17,60–62 On the basis of large published data, microvolt TWA18–20,63–75 and LGE-CMR21–28,76–85 seem to be the best candidates to improve risk stratification for SD in DCM.
Microvolt T-wave alternans
The mechanism relating TWA to the risk of SD in DCM is probably related to the ability of TWA to provide a quantitative assessment of temporal and spatial heterogeneity of repolarization which facilitates ventricular arrhythmias. Differently from LVEF which only indirectly probes electrophysiological substrate, TWA is directly linked to cellular arrhythmia mechanisms arising from calcium cycling.63,64 A review of the published data revealed seven prospective studies which have included >100 patients with DCM and an arrhythmic endpoint (pooled number of patients 1746).65–71 All studies (except two studies which tested performance in the absence of a beta-blocker66,70) documented the ability of TWA to predict the risk of SD, with a high negative predictive value (Table 3). In addition, seven meta-analyses18–20,72–75 examined only DCM patients or a mix of ischaemic and DCM patients (Table 4). All meta-analyses confirmed the arrhythmic risk predictive value of TWA test without differences between the ischaemic and DCM patients. Approximately one-third of the DCM patients had a negative TWA test, and these patients had a significantly better prognosis compared with those patients who had an abnormal (positive or indeterminate) TWA test.73 The predictive value of the TWA test was higher if it was performed in the presence of a beta-blocker therapy, as confirmed by two meta-analyses.20,75
Results of large (>100 patients) prospective studies of microvolt TWA testing for arrhythmic risk stratification in patients with DCM65–71
Study . | DCM pts (n) . | Mean FU m. . | TWA on BB . | Point estimate (CI 95%) . | P-value . | NPV (%) . |
---|---|---|---|---|---|---|
Kitamura et al.65 | 104 | 21 | Yes | RR 8.8 (1.2–65.4) | 0.0001 | 97 |
Grimm et al.66 | 263 | 52 | No | RR 1.3 (0.59–2.90) | NS | 90 |
Hohnloser et al.67 | 137 | 14 | Yes | RR 3.44 (1.09–10.91) | 0.035 | 94 |
Bloomfield et al.68 | 282 | 20 | Yes | Undefined | <0.001 | 100 |
Salerno-Uriarte et al.69 | 446 | 19a | Yes | HR 5.53 (1.29–23.65) | 0.004 | 98 |
Gold et al.70 | 250 | 30a | No | HR 1.67 (0.70–3.99) | NS | 88 |
Shizuta et al.71 | 264 | 36a | Yes | Undefined | 0.047 | 100 |
Total | 1746 |
Study . | DCM pts (n) . | Mean FU m. . | TWA on BB . | Point estimate (CI 95%) . | P-value . | NPV (%) . |
---|---|---|---|---|---|---|
Kitamura et al.65 | 104 | 21 | Yes | RR 8.8 (1.2–65.4) | 0.0001 | 97 |
Grimm et al.66 | 263 | 52 | No | RR 1.3 (0.59–2.90) | NS | 90 |
Hohnloser et al.67 | 137 | 14 | Yes | RR 3.44 (1.09–10.91) | 0.035 | 94 |
Bloomfield et al.68 | 282 | 20 | Yes | Undefined | <0.001 | 100 |
Salerno-Uriarte et al.69 | 446 | 19a | Yes | HR 5.53 (1.29–23.65) | 0.004 | 98 |
Gold et al.70 | 250 | 30a | No | HR 1.67 (0.70–3.99) | NS | 88 |
Shizuta et al.71 | 264 | 36a | Yes | Undefined | 0.047 | 100 |
Total | 1746 |
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; TWA, microvolt T-wave alternans; BB, beta-blockers; CI, confidence interval; NPV, negative predictive value; RR, relative risk; HR, hazard ratio; Undefined, hazard ratio undefined as there were no events in TWA negative group.
aMedian FU months.
Results of large (>100 patients) prospective studies of microvolt TWA testing for arrhythmic risk stratification in patients with DCM65–71
Study . | DCM pts (n) . | Mean FU m. . | TWA on BB . | Point estimate (CI 95%) . | P-value . | NPV (%) . |
---|---|---|---|---|---|---|
Kitamura et al.65 | 104 | 21 | Yes | RR 8.8 (1.2–65.4) | 0.0001 | 97 |
Grimm et al.66 | 263 | 52 | No | RR 1.3 (0.59–2.90) | NS | 90 |
Hohnloser et al.67 | 137 | 14 | Yes | RR 3.44 (1.09–10.91) | 0.035 | 94 |
Bloomfield et al.68 | 282 | 20 | Yes | Undefined | <0.001 | 100 |
Salerno-Uriarte et al.69 | 446 | 19a | Yes | HR 5.53 (1.29–23.65) | 0.004 | 98 |
Gold et al.70 | 250 | 30a | No | HR 1.67 (0.70–3.99) | NS | 88 |
Shizuta et al.71 | 264 | 36a | Yes | Undefined | 0.047 | 100 |
Total | 1746 |
Study . | DCM pts (n) . | Mean FU m. . | TWA on BB . | Point estimate (CI 95%) . | P-value . | NPV (%) . |
---|---|---|---|---|---|---|
Kitamura et al.65 | 104 | 21 | Yes | RR 8.8 (1.2–65.4) | 0.0001 | 97 |
Grimm et al.66 | 263 | 52 | No | RR 1.3 (0.59–2.90) | NS | 90 |
Hohnloser et al.67 | 137 | 14 | Yes | RR 3.44 (1.09–10.91) | 0.035 | 94 |
Bloomfield et al.68 | 282 | 20 | Yes | Undefined | <0.001 | 100 |
Salerno-Uriarte et al.69 | 446 | 19a | Yes | HR 5.53 (1.29–23.65) | 0.004 | 98 |
Gold et al.70 | 250 | 30a | No | HR 1.67 (0.70–3.99) | NS | 88 |
Shizuta et al.71 | 264 | 36a | Yes | Undefined | 0.047 | 100 |
Total | 1746 |
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; TWA, microvolt T-wave alternans; BB, beta-blockers; CI, confidence interval; NPV, negative predictive value; RR, relative risk; HR, hazard ratio; Undefined, hazard ratio undefined as there were no events in TWA negative group.
aMedian FU months.
Meta-analysis . | Total patients (n) . | DCM patients (n) . | RR (95% CI) . | NPV% . |
---|---|---|---|---|
Studies with only non-ischaemic dilated cardiomyopathy patients | ||||
De Ferrari and Sanzo73 | 1456 | 1456 | 2.99 (1.88–4.75) | 96 |
Studies with mixed ischaemic and non-ischaemic dilated cardiomyopathy patients | ||||
Gehi et al.72 | 2608 | 632 | 3.67 (1.50–8.96) | 95 |
van der Avoort et al.74 | 1946 | 658 | 2.6 (1.4–5.8)a | 99a |
Merchant et al.19 | 2883 | 566 | – | 98a |
Gupta et al.18 | 5945 | 1682 | 3.68 (2.23–6.07)a | 95a |
Studies with TWA when beta-blockers were administered | ||||
Chan et al.75 | 1277 | 865 | 5.39 (2.68–10.8)a | 98a |
Calò et al.20 | 2488 | 865 | 5.88 (3.38–10.23)a | 98a |
Studies with TWA when beta-blockers were withheld | ||||
Chan et al.75 | 2662 | 513 | 1.40 (1.06–1.84)a | 91a |
Calò et al.20 | 2662 | 513 | 1.63 (1.30–2.04)a | 90a |
Meta-analysis . | Total patients (n) . | DCM patients (n) . | RR (95% CI) . | NPV% . |
---|---|---|---|---|
Studies with only non-ischaemic dilated cardiomyopathy patients | ||||
De Ferrari and Sanzo73 | 1456 | 1456 | 2.99 (1.88–4.75) | 96 |
Studies with mixed ischaemic and non-ischaemic dilated cardiomyopathy patients | ||||
Gehi et al.72 | 2608 | 632 | 3.67 (1.50–8.96) | 95 |
van der Avoort et al.74 | 1946 | 658 | 2.6 (1.4–5.8)a | 99a |
Merchant et al.19 | 2883 | 566 | – | 98a |
Gupta et al.18 | 5945 | 1682 | 3.68 (2.23–6.07)a | 95a |
Studies with TWA when beta-blockers were administered | ||||
Chan et al.75 | 1277 | 865 | 5.39 (2.68–10.8)a | 98a |
Calò et al.20 | 2488 | 865 | 5.88 (3.38–10.23)a | 98a |
Studies with TWA when beta-blockers were withheld | ||||
Chan et al.75 | 2662 | 513 | 1.40 (1.06–1.84)a | 91a |
Calò et al.20 | 2662 | 513 | 1.63 (1.30–2.04)a | 90a |
The endpoints were arrhythmic events and SD in all studies and appropriate ICD discharges, cardiac death and all-cause mortality in some studies.
DCM, non-ischaemic dilated cardiomyopathy; RR, relative risk; CI, confidence interval; NPV, negative predictive value.
aValues of a mix of ischaemic and DCM patients.
Meta-analysis . | Total patients (n) . | DCM patients (n) . | RR (95% CI) . | NPV% . |
---|---|---|---|---|
Studies with only non-ischaemic dilated cardiomyopathy patients | ||||
De Ferrari and Sanzo73 | 1456 | 1456 | 2.99 (1.88–4.75) | 96 |
Studies with mixed ischaemic and non-ischaemic dilated cardiomyopathy patients | ||||
Gehi et al.72 | 2608 | 632 | 3.67 (1.50–8.96) | 95 |
van der Avoort et al.74 | 1946 | 658 | 2.6 (1.4–5.8)a | 99a |
Merchant et al.19 | 2883 | 566 | – | 98a |
Gupta et al.18 | 5945 | 1682 | 3.68 (2.23–6.07)a | 95a |
Studies with TWA when beta-blockers were administered | ||||
Chan et al.75 | 1277 | 865 | 5.39 (2.68–10.8)a | 98a |
Calò et al.20 | 2488 | 865 | 5.88 (3.38–10.23)a | 98a |
Studies with TWA when beta-blockers were withheld | ||||
Chan et al.75 | 2662 | 513 | 1.40 (1.06–1.84)a | 91a |
Calò et al.20 | 2662 | 513 | 1.63 (1.30–2.04)a | 90a |
Meta-analysis . | Total patients (n) . | DCM patients (n) . | RR (95% CI) . | NPV% . |
---|---|---|---|---|
Studies with only non-ischaemic dilated cardiomyopathy patients | ||||
De Ferrari and Sanzo73 | 1456 | 1456 | 2.99 (1.88–4.75) | 96 |
Studies with mixed ischaemic and non-ischaemic dilated cardiomyopathy patients | ||||
Gehi et al.72 | 2608 | 632 | 3.67 (1.50–8.96) | 95 |
van der Avoort et al.74 | 1946 | 658 | 2.6 (1.4–5.8)a | 99a |
Merchant et al.19 | 2883 | 566 | – | 98a |
Gupta et al.18 | 5945 | 1682 | 3.68 (2.23–6.07)a | 95a |
Studies with TWA when beta-blockers were administered | ||||
Chan et al.75 | 1277 | 865 | 5.39 (2.68–10.8)a | 98a |
Calò et al.20 | 2488 | 865 | 5.88 (3.38–10.23)a | 98a |
Studies with TWA when beta-blockers were withheld | ||||
Chan et al.75 | 2662 | 513 | 1.40 (1.06–1.84)a | 91a |
Calò et al.20 | 2662 | 513 | 1.63 (1.30–2.04)a | 90a |
The endpoints were arrhythmic events and SD in all studies and appropriate ICD discharges, cardiac death and all-cause mortality in some studies.
DCM, non-ischaemic dilated cardiomyopathy; RR, relative risk; CI, confidence interval; NPV, negative predictive value.
aValues of a mix of ischaemic and DCM patients.
Gupta et al.18 performed a meta-analysis of 5945 patients, to determine the ability of TWA to modify risk assessment of ventricular tachyarrhythmia events (VTE) and SD across a series of patient risk profiles using likelihood ratio (LR) testing. In SCD-HeFT-type patients (n = 2036), the annualized risks of VTE were 5.91%. T-wave alternans results (abnormal or negative) would divide these subjects into a high-risk group with a 7.7% and a low-risk group with a 2.6% risk, with an absolute risk difference in reclassification of 5.1%. Despite the fact that the LR value for TWA negative test was 0.44 (which is considered small to change clinical decisions) the post-test levels of annualized risk in this low-risk group both for VTE (2.6%) and SD (1.3%) were very low. In agreement with these results, in a pooled cohort analysis (2883 patients) published by Merchant et al.,19 the annual SD event rate was 0.9% in the group of patients with a negative TWA and LVEF ≤ 35%. Taking into account these very low SD risk values together with the ICD adverse effects, it is very unlikely that these patients would benefit from ICD therapy.
The Consensus Guideline by the International Society for Holter and Non-invasive Electrocardiology, published in 2011,64 states that TWA provides valuable information concerning the risk of cardiovascular mortality and SD, beyond standard clinical variables for cardiovascular diseases. As limitations, TWA should not be used as a sole parameter either to rule in or rule out ICD therapy, but it should be included in an algorithm with LVEF (Figure 1). Moreover, TWA test should be standardized and performed on discharge of beta-blocker. Finally, there is no randomized trial which confirms the ability of the TWA test to identify DCM patients at low risk of SD among those with LVEF ≤ 35%.
Cardiac magnetic resonance with late gadolinium enhancement
The remodelling process in DCM patients is characterized by changes in extracellular matrix, including fibrosis formation.76 The presence and extent of myocardial tissue heterogeneity with scars and interstitial fibrosis provides a substrate for ventricular arrhythmias which is considered an important cause of SD in DCM.77–79 Fibrosis can be investigated by either testing serum biomarkers or performing LGE-CMR. A few studies on serum markers of collagen turnover predicted future shocks in ICD recipients with DCM on OMT.80,81 More data are available on LGE-CMR which demonstrate myocardial scar with proven histopathological correlation.82
Assumull et al.83 were the first to observe that LGE-CMR midwall fibrosis was present in many patients with DCM and was predictive of SD/VT (P = 0.03). A review of the published data revealed 10 prospective studies in DCM patients (with a pooled total of 1405 patients) with identifiable arrhythmic endpoints (Table 5).21–27,83–85 All studies, but one,85 showed direct correlation between the presence and the degree of myocardial scar and arrhythmic events, with a high negative predictive value. The annualized risk of arrhythmic events was very low in patients without myocardial fibrosis, ranging from 0 to 3.04% (in 9 of 10 studies the value was <3%). The endpoints included appropriate ICD therapy in 8 of 10 studies and it is important to realize that ICD shock, and certainly antitachycardia pacing, cannot be considered 1 : 1 as a prevented SD.
Study . | DCM patients (n) . | Arrhythmic endpoint . | Mean FU m. . | Scar prediction of AE (CI 95%) . | P-value . | −LGE, year AE risk (%) . | NPV (%) . |
---|---|---|---|---|---|---|---|
Assomull et al.83 | 101 | SD, VT | 22 | +LGE: HR 5.2 (1.0–26.9) | 0.03 | 1.68 | 95 |
Iles et al.84 | 61 | ICD therapy | 24a | +LGE vs. −LGE: 29% vs. 0% | <0.01 | 0 | 100 |
Lehrke et al.85 | 184 | ICD therapy | 22 | +LGE vs. −LGE: 8.3% vs. 1.8% | 0.06 | 1.48 | 97 |
Klem et al.21 | 64 | ICD therapy, TD | 24a | +LGE: HR 4.71 (1.02–21.8) | 0.05 | 2.8 | 93 |
Wu et al.22 | 98 | ICD therapy, CD | 42a | Grey zone: HR 4.6 (1.4–15.4) | 0.01 | 0.7b | 96 |
Gao et al.23 | 65 | ICD therapy, SD, a.SD | 21 | Total scar: HR 1.39 (1.11–1.74) | 0.004 | 3.04 | 86 |
Fernandez-Armenta et al.24 | 78 | ICD therapy | 25 | Scar mass ≥16% vs. <16%: 11.5% vs. 0% | <0.01 | 0 | 100 |
Leyva et al.25 | 97 | SD | 34a | +LGE vs. −LGE: 15% vs. 0% | 0.0029 | 0 | 100 |
Muller et al.26 | 185 | ICD therapy, a.SD | 21a | +LGE vs. −LGE: 17% vs. 4% | 0.006 | 2.52 | 95 |
Gulati et al.27 | 472 | ICD therapy., SD, a.SD | 63a | +LGE: HR 5.24 (3.15–8.72) | <0.001 | 1.31 | 93 |
Total | 1405 |
Study . | DCM patients (n) . | Arrhythmic endpoint . | Mean FU m. . | Scar prediction of AE (CI 95%) . | P-value . | −LGE, year AE risk (%) . | NPV (%) . |
---|---|---|---|---|---|---|---|
Assomull et al.83 | 101 | SD, VT | 22 | +LGE: HR 5.2 (1.0–26.9) | 0.03 | 1.68 | 95 |
Iles et al.84 | 61 | ICD therapy | 24a | +LGE vs. −LGE: 29% vs. 0% | <0.01 | 0 | 100 |
Lehrke et al.85 | 184 | ICD therapy | 22 | +LGE vs. −LGE: 8.3% vs. 1.8% | 0.06 | 1.48 | 97 |
Klem et al.21 | 64 | ICD therapy, TD | 24a | +LGE: HR 4.71 (1.02–21.8) | 0.05 | 2.8 | 93 |
Wu et al.22 | 98 | ICD therapy, CD | 42a | Grey zone: HR 4.6 (1.4–15.4) | 0.01 | 0.7b | 96 |
Gao et al.23 | 65 | ICD therapy, SD, a.SD | 21 | Total scar: HR 1.39 (1.11–1.74) | 0.004 | 3.04 | 86 |
Fernandez-Armenta et al.24 | 78 | ICD therapy | 25 | Scar mass ≥16% vs. <16%: 11.5% vs. 0% | <0.01 | 0 | 100 |
Leyva et al.25 | 97 | SD | 34a | +LGE vs. −LGE: 15% vs. 0% | 0.0029 | 0 | 100 |
Muller et al.26 | 185 | ICD therapy, a.SD | 21a | +LGE vs. −LGE: 17% vs. 4% | 0.006 | 2.52 | 95 |
Gulati et al.27 | 472 | ICD therapy., SD, a.SD | 63a | +LGE: HR 5.24 (3.15–8.72) | <0.001 | 1.31 | 93 |
Total | 1405 |
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; AE, arrhythmic events; CI, confidence interval; HR, hazard ratio; +LGE, late gadolinium enhancement positive; −LGE, late gadolinium enhancement negative; SD, sudden death; VT, ventricular tachycardia; TD, all-cause mortality; CD, cardiac death; a.SD, aborted sudden death.
aMedian FU months.
bPatients in the lowest tertile for both grey zone and high-sensitivity C-reactive protein.
Study . | DCM patients (n) . | Arrhythmic endpoint . | Mean FU m. . | Scar prediction of AE (CI 95%) . | P-value . | −LGE, year AE risk (%) . | NPV (%) . |
---|---|---|---|---|---|---|---|
Assomull et al.83 | 101 | SD, VT | 22 | +LGE: HR 5.2 (1.0–26.9) | 0.03 | 1.68 | 95 |
Iles et al.84 | 61 | ICD therapy | 24a | +LGE vs. −LGE: 29% vs. 0% | <0.01 | 0 | 100 |
Lehrke et al.85 | 184 | ICD therapy | 22 | +LGE vs. −LGE: 8.3% vs. 1.8% | 0.06 | 1.48 | 97 |
Klem et al.21 | 64 | ICD therapy, TD | 24a | +LGE: HR 4.71 (1.02–21.8) | 0.05 | 2.8 | 93 |
Wu et al.22 | 98 | ICD therapy, CD | 42a | Grey zone: HR 4.6 (1.4–15.4) | 0.01 | 0.7b | 96 |
Gao et al.23 | 65 | ICD therapy, SD, a.SD | 21 | Total scar: HR 1.39 (1.11–1.74) | 0.004 | 3.04 | 86 |
Fernandez-Armenta et al.24 | 78 | ICD therapy | 25 | Scar mass ≥16% vs. <16%: 11.5% vs. 0% | <0.01 | 0 | 100 |
Leyva et al.25 | 97 | SD | 34a | +LGE vs. −LGE: 15% vs. 0% | 0.0029 | 0 | 100 |
Muller et al.26 | 185 | ICD therapy, a.SD | 21a | +LGE vs. −LGE: 17% vs. 4% | 0.006 | 2.52 | 95 |
Gulati et al.27 | 472 | ICD therapy., SD, a.SD | 63a | +LGE: HR 5.24 (3.15–8.72) | <0.001 | 1.31 | 93 |
Total | 1405 |
Study . | DCM patients (n) . | Arrhythmic endpoint . | Mean FU m. . | Scar prediction of AE (CI 95%) . | P-value . | −LGE, year AE risk (%) . | NPV (%) . |
---|---|---|---|---|---|---|---|
Assomull et al.83 | 101 | SD, VT | 22 | +LGE: HR 5.2 (1.0–26.9) | 0.03 | 1.68 | 95 |
Iles et al.84 | 61 | ICD therapy | 24a | +LGE vs. −LGE: 29% vs. 0% | <0.01 | 0 | 100 |
Lehrke et al.85 | 184 | ICD therapy | 22 | +LGE vs. −LGE: 8.3% vs. 1.8% | 0.06 | 1.48 | 97 |
Klem et al.21 | 64 | ICD therapy, TD | 24a | +LGE: HR 4.71 (1.02–21.8) | 0.05 | 2.8 | 93 |
Wu et al.22 | 98 | ICD therapy, CD | 42a | Grey zone: HR 4.6 (1.4–15.4) | 0.01 | 0.7b | 96 |
Gao et al.23 | 65 | ICD therapy, SD, a.SD | 21 | Total scar: HR 1.39 (1.11–1.74) | 0.004 | 3.04 | 86 |
Fernandez-Armenta et al.24 | 78 | ICD therapy | 25 | Scar mass ≥16% vs. <16%: 11.5% vs. 0% | <0.01 | 0 | 100 |
Leyva et al.25 | 97 | SD | 34a | +LGE vs. −LGE: 15% vs. 0% | 0.0029 | 0 | 100 |
Muller et al.26 | 185 | ICD therapy, a.SD | 21a | +LGE vs. −LGE: 17% vs. 4% | 0.006 | 2.52 | 95 |
Gulati et al.27 | 472 | ICD therapy., SD, a.SD | 63a | +LGE: HR 5.24 (3.15–8.72) | <0.001 | 1.31 | 93 |
Total | 1405 |
DCM, non-ischaemic dilated cardiomyopathy; FU m., follow-up months; AE, arrhythmic events; CI, confidence interval; HR, hazard ratio; +LGE, late gadolinium enhancement positive; −LGE, late gadolinium enhancement negative; SD, sudden death; VT, ventricular tachycardia; TD, all-cause mortality; CD, cardiac death; a.SD, aborted sudden death.
aMedian FU months.
bPatients in the lowest tertile for both grey zone and high-sensitivity C-reactive protein.
Klem et al.21 reported that myocardial scarring, as detected by CMR, was an independent predictor of death or appropriate ICD discharge in 137 patients (64 with DCM) who were considered for ICD implantation; in patients with LVEF ≤ 30%, minimal or no scarring identified a low-risk cohort, similar to patients with LVEF > 30%, suggesting that the presence of a ventricular scar may be as powerful a prognostic marker as LVEF. However, the majority of these studies were limited by small sample size, and by different study designs and methodologies. Only one meta-analysis28 examined a mix of ischaemic and DCM patients (1105 patients of which 491 with DCM) and confirmed the ability of LGE-CMR to predict the risk of ventricular tachyarrhythmias (RR 4.33; 95% CI, 2.98–6.29; P < 0.00001).
Very recently, Gulati et al.27 published a large prospective study of 472 patients with DCM who were followed up for a median duration of 5.3 years. The arrhythmic composite endpoint (SD, appropriate ICD shock, and non-fatal VT/VF) was reached by 29.6% of patients with LGE-CMR midwall fibrosis vs. 7.0% (1.31% per year) of patients without fibrosis (HR: 5.24; 95% CI: 3.15–8.72; absolute risk difference 22.6%; P < 0.001). The data were confirmed by multivariable analysis. For the composite arrhythmic endpoint, the addition of midwall fibrosis status to a risk model based on LVEF significantly improved the reclassification of patients as high or low arrhythmic risk. Overall, 29% of patients were correctly reclassified (net reclassification improvement, 0.29; 95% CI: 0.11–0.48; P = 0.002).
On the basis of these data we believe that LGE-CMR contributes to stratify the risk of cardiovascular mortality and SD, particularly if included in an algorithm with LVEF (Figure 1). In the absence of midwall fibrosis, the risk of SD seems to be very low, and it is very unlikely that these patients would benefit from ICD therapy. A standardization of LGE-CMR evaluation and reporting could help. In fact, in some studies scar size,21 scar mass percentage,24 total scar,23 amount of heterogeneous myocardial tissue and dense core scar,22 and the border zone percentage of scar24 analyses had a better diagnostic accuracy than the simple description of the presence or absence of midwall fibrosis. Finally, as for TWA, there is no randomized study which confirms the ability of LGE-CMR to identify DCM patients at low risk of SD among those with LVEF ≤ 35%.
Algorithm
Recent data suggest that because of the complexities of the substrates underlying SD, multiple risk factors used in combination are likely to provide better prediction of SD risk than any individual risk marker.17 In Figure 1, we report an algorithm for selection of DCM patients with ICD indication in SD primary prevention. In the group of patients with LVEF ≤ 35% on OMT and no important comorbidities, about one-third have a negative TWA test or absence of midwall fibrosis at LGE-CMR. Among this one-third of patients, there is a group with familial DCM without SD in relatives and without Lamin A/C mutation. Cumulatively, the algorithm could select a group of patients (with negative TWA or absence of midwall fibrosis, no SD family history, and absence of Lamin A/C mutation) at very low risk of SD, despite a LVEF ≤ 35%. It is very unlikely that these patients benefit from ICD therapy. However, the proposed algorithm has been generated taking into account scientific data robust enough to either be considered in existing guidelines64 or to be replicated in multiple studies. Although it has not been validated by any randomized trial, it may represent a flexible starting point to reconsider clinical indication to ICD implantation in DCM patients.
Conclusions
The selection of patients according to current guidelines shows that most DCM patients actually do not benefit from ICD implantation3,4 and may suffer collateral effects. In addition, many patients who are at risk of SD are not identified by the guidelines, because a large proportion of SD patients exhibit only mildly depressed LVEF.15,16,44 Identifying high-risk patients on the sole basis of LVEF does not allow the best use of the benefits of ICD therapy.9
In clinical practice, decision-making should be based upon simple criteria, but the lack of a highly sensitive and specific predictive marker calls for a more complex evaluation. SD risk stratification could improve by adding other risk markers, such as TWA, LGE-CMR, and genetics tests to LVEF. For example, when following the algorithm suggested in Figure 1, one-third of patients with ‘guidelines indication’ to ICD would be reassigned to a lower-risk group which is unlikely to significantly benefit from ICD therapy. Other or new non-invasive and invasive risk markers could be included in different algorithms. What is important is to change the approach to SD risk stratification from using only LVEF values to utilizing a multiple risk markers model. In addition, a report from the National Heart, Lung, and Blood Institute and the Heart Rhythm Society Workshop17 emphasized that multiple risk factors used in combination are likely to provide better prediction of SD risk than any individual risk marker because of the complexities of the substrates underlying SD.
Given the undiscovered individual risk and the costs, we need trials including novel clinical and genetic risk markers to improve risk stratification and increase the efficacy of ICD therapy. Unfortunately, to the best of our knowledge, only one of the ongoing trials86 focuses on SD multiple risk marker stratification in DCM patients, and its conclusion is expected in 2014. Waiting for this and other possible trials results, we feel that the amount of data on risk markers different from LVEF, could already justify a modification in patient selection.
Conflict of interest: None declared.
Funding
This work was supported by a 2010 Grant from Fondazione Cassa di Risparmio di Trento e Rovereto (Caritro), Trento, Italy, and Grants European Union INHERITANCE project 241924, Health-2009-2.4.2-3.