Author + information
- Gregory M. Marcus, MD, MAS* ()
- ↵*Reprint requests and correspondence:
Dr. Gregory M. Marcus, Division of Cardiology, Electrophysiology Section, University of California, 500 Parnassus Avenue, MUE 434, San Francisco, California 94143-1354
There is general agreement that left ventricular ejection fraction (EF) remains the best available predictor in helping to select appropriate patients for a primary prevention implantable cardioverter-defibrillator (ICD) (1). However, it is also widely recognized that this measurement alone is suboptimal, lacking in both sensitivity and specificity (2). In particular, primum non nocere (“first, do no harm”) becomes a particularly relevant consideration when we face patients suffering from the complications of an ICD implantation with the knowledge that they might never need the device. This issue is relevant not only to patients that are candidates for their initial primary prevention ICD but increasingly to those ICD patients who have never exhibited a malignant ventricular arrhythmia presenting for a second, third, or fourth procedure because of a lead fracture, device recall, or generator at the elective replacement interval. Better predictors of incident malignant ventricular arrhythmias are needed.
These issues are particularly pertinent to patients with nonischemic dilated cardiomyopathy (NIDC) for several reasons. First, the data supporting primary prevention ICDs in this population are not as robust as the data for those with a reduced EF due to coronary disease (1,3–6). Second, tools used to help with risk stratification, such as invasive electrophysiology testing, are likely not as useful in patients with NIDC (7). Finally, because many of the youngest patients meeting guideline criteria for primary prevention ICDs will more often have a nonischemic than an ischemic etiology, NIDC patients might more often have the most to gain from a new clinically useful predictor of malignant ventricular arrhythmias.
Much research has sought to determine useful tests for sudden arrhythmic death risk stratification. Whether based on electrophysiologic parameters (such as QRS duration, invasive electrophysiology testing, the signal averaged electrocardiogram, and micro-volt T wave alternans) or autonomic parameters (such as heart rate variability, heart rate turbulence, or baroreflex sensitivity), none have proven to provide widely applicable accuracy (7,8). Recently, methods to better understand the structural substrate have been investigated. Interesting data suggest that the nature of the ventricular scar as determined by magnetic resonance imaging might reveal those hearts at particularly high risk, even in those with NIDC (9,10). Because cardiac magnetic resonance imaging might not be a feasible screening tool for primary prevention ICD patients, the search for less expensive and more easily obtained measures of arrhythmogenic myocardial scar has begun.
In this issue of the Journal, Kanoupakis et al. (11) describe their study investigating the utility of several markers of fibrosis, C-terminal propeptide of collagen type-1, C-terminal telopeptide of collagen type-1, matrix metalloproteinase-1, and the tissue inhibitor of matrix metalloproteinase-1, as predictors of appropriate ICD therapies in 70 NIDC patients. These serologic markers reflect the synthesis and degradation of type-1 collagen, the most abundant form of collagen in myocardium. The underlying hypothesis is that a snapshot view of these peripheral serum levels would provide a window into understanding the arrhythmic potential of myocardial scar.
Fourteen of the 70 patients experienced at least 1 appropriate ICD therapy over 1 year of follow-up. Although no differences in demographic data, associated medical conditions, or medications between those with and without the outcome were observed, 3 of the 4 serologic markers were significantly higher in those with an appropriate ICD therapy than in those without. These differences were not only statistically significant but also exhibited a potentially clinically useful separation in distributions, particularly for C-terminal telopeptide of collagen type-1. This appropriately motivated an assessment of receiver-operating characteristics curves, with a suggestion that these markers might exhibit excellent test characteristics.
In addition to addressing an important topic and providing early research toward a potentially new method in risk assessment, this study has several strengths: the population is quite homogenous, comprised of only NIDC patients. Three markers exhibited similar patterns, providing consistency in support of their hypothesis. There was also no apparent loss to follow-up, and the outcome was based on objective and measurable data. However, before these findings are considered for integration into clinical practice or even further study, it is important to highlight several important limitations.
First, the outcome was appropriate ICD therapy, an imperfect surrogate for mortality. In addition, the primary outcome was heterogeneous, comprising either antitachycardia pacing or ICD shock. Although most patients with a primary outcome experienced an ICD shock, arguably a clinically relevant outcome given the associated severe discomfort, some of the antitachycardia pacing events might have represented treatment for ventricular tachycardia that would have spontaneously resolved.
Second, as is the case with essentially any observational study, residual confounding must always be considered as an explanation for positive findings. The groups seemed to be well-balanced by all measured clinical characteristics, but a lack of statistical significance between such characteristics does not eliminate the possibility of confounding. In addition, the variables assessed were not comprehensive. For example, body mass index was not reported. Could it be that these markers are increased in obesity and that obesity was the true predictor of ICD therapy? To address unmeasured and unknown confounders, a randomized study is required.
Third, external validity or generalizability to our patients in practice deserves consideration. All patients had evidence of nonsustained ventricular tachycardia on Holter study or inducible ventricular tachycardia with invasive electrophysiology testing. Per the most recent guidelines (1), based primarily on SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial) (6), NIDC patients can be appropriate candidates for primary prevention ICDs without any prior evidence of ventricular arrhythmias. The mechanism of ventricular arrhythmias might be different in those with prior evidence of nonsustained or inducible ventricular tachycardia compared with those with no known antecedent arrhythmias, potentially making measurement of these markers less relevant in many NIDC patients. In addition, this study was conducted in Crete. Perhaps the common etiologies of NIDC in that single location are substantially different than elsewhere, and maybe the predictive ability of these fibrotic markers is uniquely conducive to NIDC of those particular etiologies. A larger study enrolling a more diverse patient population would be needed to teach us whether these same markers predict arrhythmias in patients such as the former methamphetamine user, the patient with adriamycin toxicity, the patient with a familial cardiomyopathy, or the hypertensive African American.
One apparent limitation involves the relatively small sample size. However, it is important to emphasize that this should not increase the risk of a type I error. In other words, a small sample size or small number of outcomes is not responsible for spurious statistically significant associations. However, it does increase the likelihood of a type II error (i.e., false negatives or insufficient power to detect a true difference). In fact, the relatively small numbers might explain why clinical characteristics expected to be predictive of the outcome, such as EF and New York Heart Association functional class, were not significantly associated with ICD therapies.
In conclusion, although there is insufficient evidence to consider measurement of serologic markers of fibrosis as useful predictors of appropriate ICD therapies in NIDC patients in clinical practice, the intriguing findings described in this report invite a potentially fruitful new line of investigation. Ancillary studies of completed prospective randomized trials with available baseline serum would efficiently leverage existing resources and potentially validate these findings in a more definitive fashion.
Dr. Marcus has received research support and speaker fees from St. Jude Medical and Biotronik.
↵* Editorials published in the Journal of the American College of Cardiologyreflect the views of the authors and do not necessarily represent the views of JACCor the American College of Cardiology.
- American College of Cardiology Foundation
- Epstein A.E.,
- DiMarco J.P.,
- Ellenbogen K.A.,
- et al.
- Tung R.,
- Zimetbaum P.,
- Josephson M.E.
- Strickberger S.A.,
- Hummel J.D.,
- Bartlett T.G.,
- et al.
- Bansch D.,
- Antz M.,
- Boczor S.,
- et al.
- Assomull R.G.,
- Prasad S.K.,
- Lyne J.,
- et al.
- Nazarian S.,
- Bluemke D.A.,
- Lardo A.C.,
- et al.
- Kanoupakis E.M.,
- Manios E.G.,
- Kallergis E.M.,
- et al.