Author + information
- ↵⁎Reprint requests and correspondence:
Dr. Gregory K. Feld, 4168 Front Street, San Diego, California 92103-8649
In this issue of the Journal, Costantini et al. (1) publish a study comparing noninvasive microvolt T-wave alternans (MTWA) with invasive electrophysiological study with ventricular programmed stimulation (EPS) to identify patients at risk of sudden cardiac death (SCD) who might benefit from an implantable cardioverter-defibrillator (ICD).
The authors (1) noted that primary prevention trials, for example MADIT-I (Multicenter Automatic Defibrillator Trial) and MUSTT (Multicenter Unsustained Tachycardia Trial) (2,3), used EPS to identify high risk of SCD in patients with prior myocardial infarction, reduced left ventricular ejection fraction (LVEF) of ≤0.35 to 0.40, spontaneous nonsustained ventricular tachycardia (NSVT), and inducible sustained ventricular tachycardia (VT) at EPS (which was not suppressed by antiarrhythmic drugs in MADIT-I). This rigorous invasive approach identified patients in whom overall mortality was significantly reduced by ICD implantation compared with conventional antiarrhythmic drug therapy during maximum follow-up of 5 years. Of note, overall mortality was also high (44%) in the MUSTT registry in patients who had no inducible VT at EPS (4). Thus, in patients with myocardial infarction and reduced LVEF, the risk of mortality is high regardless of inducibility of VT at EPS, although EPS might identify patients at highest risk who are more likely to benefit from ICD implantation.
In the subsequent MADIT-II trial (5), designed to eliminate the need for EPS, patients with myocardial infarction and LVEF ≤0.30 were randomized to ICD implantation versus conventional medical therapy only. ICD implantation produced a smaller yet significant reduction in overall mortality versus conventional medical therapy (14.2% vs. 19.8%) during a maximum follow-up of 53 months.
Among alternative noninvasive screening tools, MTWA has been shown to be useful for risk stratification for primary prevention of SCD (6–9). However, its high negative predictive value (NPV) might be its most valuable asset (6–9). Therefore, Costantini et al. (1) designed the ABCD trial to test the hypothesis that, in patients with coronary artery disease and reduced LVEF, MTWA would perform at least as well as EPS to identify increased risk for SCD. In addition, the authors hypothesized that MTWA, either alone or in combination with EPS, would identify patients more likely to benefit from ICD implantation compared with stratification by LVEF alone.
Costantini et al. (1) partly justify their study by noting that 4 ICDs must be implanted to save 1 life with MADIT-I and MUSTT criteria and 15 to 17 ICDs must be implanted to save 1 life with MADIT-II criteria (2–5). Moreover, many ICD patients never receive therapy. Nonetheless, current guidelines reflect MADIT-II findings (5), recommending ICD implantation for primary prevention of SCD in patients with an LVEF ≤0.35 without requiring EPS to guide therapy. The authors (1) hoped that use of MTWA might reduce this ratio of ICD implants to life saved and avoid invasive EPS.
In the ABCD trial, Costantini et al. (1) enrolled patients with NSVT and LVEF ≤0.40 attributable to ischemic heart disease. The trial was designed to compare an MTWA-guided with EPS-guided strategy in predicting arrhythmic events. The MTWA strategy was defined as positive (high risk) if MTWA was abnormal or if MTWA was indeterminate and EPS was abnormal, whereas the MTWA strategy was defined as negative (low risk) if MTWA was normal or if MTWA was indeterminate and EPS was normal. Thus, all patients with reduced LVEF would be screened noninvasively with MTWA and undergo EPS only if MTWA was indeterminate. However, because an overlap would then exist between the MTWA and EPS groups, pre-specified secondary analyses were performed with a standard definition of MTWA positivity that excludes patients with indeterminate results and an alternative validated method (9) defining patients with positive or indeterminate MTWA as abnormal and patients with negative MTWA as normal. The goal was to demonstrate noninferiority of MTWA compared with EPS in identifying patients at high risk of SCD who would benefit from ICD implantation.
Costantini et al. (1) followed 566 patients up to 2 years, 46% with positive and 25% with indeterminate MTWA and 40% with positive EPS. The ICDs were inserted in 87% overall and in 97% with positive MTWA and EPS. One-year and 2-year event rates of 7.5% and 14%, respectively, occurred in 65 patients, including 55 ICD therapies and 10 SCDs. The authors observed a 1.6% difference (9.5% for MTWA vs. 11.1% for EPS) in positive predictive value (PPV) and a 0.2% difference (95.1% for EPS vs. 95.3% for MTWA) in NPV between MTWA and EPS at 1 year, reportedly falling within their 10% definition of non-inferiority. In addition, pre-specified analyses comparing positive/negative MTWA with EPS and abnormal/normal MTWA with EPS also fell within their 10% definition of non-inferiority. Interestingly, the event rate in patients with 2 normal tests was approximately 3-fold lower than in patients with 1 abnormal test and approximately 6-fold lower than in patients with 2 abnormal tests, suggesting that MTWA and EPS were complementary in predicting outcomes. The data also showed that EPS was a significant predictor of events starting at 9 months and for the remainder of the study, whereas MTWA was predictive of clinical outcomes at 6 months but not after 12 months.
From a statistical perspective, this noninferiority analysis should be understood for what it actually shows. Conceptually, non-inferiority tests are designed to show that 1 (usually new) method is not a lot worse than another (usually established) method. The key is in setting the noninferiority margin to prospectively define what “not a lot worse” actually means. For example, if an existing treatment is 80% effective and a noninferiority margin of 10% had been pre-specified, then a significant result would show that the new treatment is better than 70% effective. Thus, the strength of the reasonable equivalence claim depends largely on the size of the pre-specified noninferiority margin.
Because, in the Costantini et al. (1) study, EPS achieved a PPV of 11.1%, the MWTA test would only need to achieve a PPV significantly better than 1.1% to demonstrate non-inferiority with a pre-specified margin of 10%. Thus, demonstrating that the difference in PPV for EPS versus MTWA is <10% is not sufficient to show that the methods are reasonably equivalent. In contrast, if a smaller margin was used, such as 2% instead of 10%, it is unlikely that the current sample size would be sufficient to provide adequate power to test the noninferiority hypothesis. The non-inferiority test for NPV is also plagued by the wide margin issue, although to a lesser degree.
Therefore, the Costantini et al. (1) data should not be taken as evidence that MTWA can replace EPS or as proof that the 2 are reasonably equivalent. What then does the study tell us? The current data are best taken as suggesting that MTWA should be studied further to determine its role in identification of patients with ischemic heart disease and reduced LVEF who are at highest risk for SCD and in whom ICD implantation might be warranted. The data also show that the event rates are low and the PPV of both methods is low. Thus, if either method is used to guide practice, the number of ICDs inserted that actually benefit patients is expected to remain low.
From a clinical perspective, the Costantini et al. study (1) also has several limitations. This study (1) demonstrated a very low event rate at 1 year (i.e., strong NPV) in the negative MTWA (95.3%) and EPS (95.2%) patients, which contrasts with the relatively high event rate (44%) in the EPS negative patients in the MUSTT registry (4). This might actually favor use of negative MTWA and EPS, alone or in combination, for stratification of patients at risk of SCD who won't benefit from ICD therapy. However, the MUSTT study (4) spanned a much longer time frame of 5 years, which might negate the NPV of MTWA suggested by this shorter-term (1.6 ± 0.6 year) study (1). It is also concerning that this study (1) showed a loss of PPV of MTWA after 12 months, compared with EPS, which predicted events after 9 months for the duration of the study, suggesting a possible difference in mechanism of SCD predicted by these 2 tests. Finally, this study (1) did not evaluate patients with nonischemic cardiomyopathy at risk of SCD, in whom ICD implantation has also been shown to significantly reduce overall mortality (10,11).
Thus, although the Costantini et al. (1) study might not have shown comparability or noninferiority of MTWA compared with EPS in identification of patients at risk of SCD who would benefit from ICD implantation, it certainly can be used as an example for the design of future similar studies. However, such a study should either incorporate a narrower margin of noninferiority or enroll a larger number of patients to ensure adequate power. It might also be appropriate in such a study to enroll patients with both ischemic and nonischemic cardiomyopathy and to follow them over a longer period of time.
Dr. Feld is a consultant to Medwaves, Inc. and Sequel Pharmaceutical Inc., and receives research and fellowship program support from Boston Scientific, Inc., Medtronic Inc., Biotronic, and St. Jude Inc.
↵⁎ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
- American College of Cardiology Foundation
- Costantini O.,
- Hohnloser S.H.,
- Kirk M.M.,
- et al.,
- ABCD Trial Investigators
- Moss A.J.,
- Hall W.J.,
- Cannom D.S.,
- et al.
- Bloomfield D.M.,
- Steinman R.C.,
- Namerow P.B.,
- et al.
- Gehi A.K.,
- Stein R.H.,
- Metz L.D.,
- Gomes J.A.
- Chow T.,
- Keriakes D.,
- Bartone C.,
- et al.
- Kaufman E.S.,
- Bloomfield D.M.,
- Steinman R.C.,
- et al.