This systematic review and meta-analysis examined MTWA as a predictor of mortality and severe arrhythmic events for primary prevention in patients with left-ventricular dysfunction. We were unable to identify any randomized trials investigating this diagnostic modality. Consequently, our conclusions are based on less rigorous cohort studies, which are further weakened by a lack of standardization. Nevertheless, a positive MTWA test result cumulatively predicted an increased risk of mortality and severe arrhythmic events compared with a negative test. MTWA may therefore possess some clinical utility as a stratification tool to assess short- to moderate-term risk in high-risk primary prevention populations. This is important as the majority of the economic burden associated with the prevention of arrhythmic events by ICD implantation occurs in this population [25, 26]. However, the wide credible interval suggests the clinical utility may range from very modest to substantial and highlights the need for additional high-quality studies to better refine these estimates. Furthermore, our review identified a dearth of data examining the utility of MTWA for long-term risk stratification; only 1 study to date has examined the ability of MTWA to predict events after 2 years of follow-up.
MTWA as a predictor of cardiac events has been examined in several non-systematic, narrative reviews [7, 8, 27–31] and one systematic review. This systematic review included a meta-analysis and also concluded that MTWA predicted an increased risk of mortality and arrhythmic events . Compared to the previous meta-analysis , which included more heterogeneous studies, we have avoided duplicate publications, added several more recent publications, and included only patient populations with significant left ventricular dysfunction. Our study is therefore more homogenous and specifically addresses the utility of MTWA in high-risk primary prevention. In addition, we have concentrated on comparing ratios of predictive values and avoided summarizing predictive values as these results are heavily dependent on study disease prevalence and follow-up time. Despite these important differences, the results of this previous study  are generally consistent with those of the present meta-analysis.
The utility of MTWA for predicting severe arrhythmic events in patients with left ventricular dysfunction also has been examined in 3 trials that have been recently presented at scientific meetings but not included in this meta-analysis. In the MASTER-I trial, which was presented at the Late Breaking Clinical Trial Session of the American Heart Association 2007 Scientific Sessions, 575 post-MI patients with left ventricular ejection fractions ≤ 30% and no history of severe arrhythmic events from 50 U.S. hospitals were followed for at least 2 years . All patients received an ICD. The primary endpoint of the MASTER I trial was life-threatening ventricular tachyarrhythmic event, defined as either arrhythmic death or appropriate ICD discharge, and the secondary endpoints were total and cause-specific mortality. Non-negative MTWA did not appear to predict life-threatening ventricular tachyarrhythmias (hazard ratio (HR): 1.26, 95% confidence interval (CI): 0.76, 2.09), although with a wide 95% CI, we are unable to rule out a clinical difference of the same magnitude as that reported in the current meta-analysis. The MASTER I investigators did find an increase in all-cause mortality among patients with non-negative MTWA (HR: 2.04, 95%CI: 1.10, 3.78). In the MASTER II trial, the utility of MTWA was examined in 303 patients with moderate ventricular dysfunction (i.e., left ventricular ejection fraction between 31% and 40%), 48% of whom received an ICD . The MASTER II trial, which excluded patients with indeterminate test results, provided inconclusive results due to a lower event rate than expected (positive MTWA vs negative MTWA: unadjusted stratified HR: 1.22, 95% CI: 0.34, 4.39; adjusted stratified HR: 1.20, 95% CI = 0.33, 4.31). In the ABCD trial, Costantini and colleagues examined the utility of MTWA and electrophysiological testing to predict different arrhythmic outcomes . This study involved 566 patients with ischemic cardiomyopathy, an ejection fraction < 40%, and documented non-sustained VT, and patients were followed for 2 years. The investigators found that, at 1 year, non-negative MTWA was an important predictor of polymorphic VT, VF, and SCD (2.7% vs 0%, p = 0.04) but not of monomorphic VT (data not provided). At 2 years, non-negative MTWA was not associated with either outcome.
Although these studies provide important information regarding the utility of MTWA, we have not included them in our meta-analysis for a number of reasons. First, they have not yet undergone the rigorous peer-review associated with publication of a full manuscript. Second, in the MASTER-I trial, all patients received an ICD regardless of MTWA test. This trial therefore examines an inherently different question than the one posed in our meta-analysis. Finally, only limited information is available for each study, rending it difficult to assess the eligibility of each study relative to the inclusion/exclusion criteria of our meta-analysis. This limited information available also prevents the inclusion of these data in the analyses, particularly in the case of the ABCD trial. Despite these limitations, it is unlikely that the inclusion of these results would alter the conclusions drawn in the present study, particularly given the consistency of the point estimates obtained in these studies with our results.
Our study has a number of strengths. This is the first systematic review and meta-analysis that focuses on the use of MTWA in the setting of primary prevention in patients with left-ventricular dysfunction. By excluding studies of healthy individuals, we provide a more valid measure of MTWA specificity and negative predictive value as it pertains to clinical practice. Second, our meta-analysis included patients with indeterminate test results, avoiding a potentially important selection bias. Third, our systematic review and meta-analysis was conducted using a pre-specified protocol and in accordance with the MOOSE criteria . Finally, this is the first meta-analysis to compare the different MTWA categories (i.e., positive vs negative, positive vs indeterminate, negative vs indeterminate), rather than simply grouping test results as negative and non-negative. Furthermore, our comparison of negative and non-negative MTWA was restricted to studies that reported indeterminate tests results, resulting in a more valid estimate of the effect of MTWA.
Our meta-analysis also has some potential limitations. First, we restricted our search to full-text studies published in English and thus may be affected by language and/or publication bias. Although our decision to not include studies recently published as abstracts may affect the precision of our estimates, it is unlikely to bias our results. These 3 studies have not been published because they are recently completed; the publication status is not due to the positive or negative nature of their results. Second, despite recommendations from the manufacturer strongly encouraging retesting following an indeterminate result , none of the studies identified in our systematic review conducted retesting. Retesting could decrease the number of patients with indeterminate test results and could potentially alter the sensitivity and specificity of MTWA. Thus, we may possibly be underestimating the potential benefits of MTWA testing. Third, there was some heterogeneity present in the individual studies included in our systematic review and meta-analysis. Studies varied in patient characteristics, duration of follow-up, study endpoints, and distribution of MTWA categories. Heterogeneity was also present in test classification, with some studies grouping patients into 3 categories (positive, negative, and indeterminate), some grouping patients into negative and non-negative, and the remaining studies excluding indeterminate tests. However, our results, although inconclusive due to wide CrIs that include both no effect and clinical important differences, suggest that there is no difference in risk of mortality or severe arrhythmic events between patients with positive MTWA and those with indeterminate results. Consequently, grouping patients as negative and non-negative is likely a valid approach. Fourth, some studies excluded indeterminate tests when calculating sensitivities and specificities. Where possible, we re-calculated the sensitivity, specificity, and positive and negative predictive values comparing non-negative and negative MTWA. However, we were unable to re-calculate these test characteristics for all studies. We also examined MTWA using the ratio of positive predictive values, which is not affected by the underlying disease prevalence. Fifth, we were limited to aggregate data and thus were unable to compare the predictive ability of MTWA to those of other tests, including New York Heart Association Class and left ventricular ejection fraction. Sixth, the ability of MTWA to predict events occurring 2 or more years after testing remains unclear. Finally, all studies included in our meta-analysis were observational studies. The potential effects of selection bias and confounding must be considered when interpreting their results.