Microvolt T-wave Alternans

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

Based on statements in the proposed decision memo and the responses by a number of experts in cardiac electrophysiology and sudden cardiac death risk stratification who have extensive experience with the Modified Moving Average (MMA) method, it seems that a number of clarifications about the methodology and supportive evidence are warranted.

We are particularly concerned that a Microvolt T-wave Alternans (MTWA) Consensus Guideline published in Journal of the American College of Cardiology (2011;58:1309–24) does not appear to have been acknowledged as pertaining to the MMA as well as to the Spectral Analytic (SA) method, as commented on by Dr. Marek Malik, who chaired the writing committee. In fact, the Consensus Guideline specified the appropriate use and standards for the MMA method and conducted a side-by-side comparison of the predictive capacity of the two techniques. The international authors, who are experts in both methods, concluded that they were equivalent in their predictive power for sudden cardiac death.

The proposed decision memo also did not acknowledge the fact that the American College of Cardiology/American Heart Association/ and European Society of Cardiology “Guidelines for Management of Patients With Ventricular Arrhythmias and the Prevention of Sudden Cardiac Death” (J Am Coll Cardiol 2006;48:e385-e484) recommended TWA monitoring on ambulatory ECG to “evaluate risk or to judge therapy” with Class I Level of Evidence A and to “improve diagnosis and risk stratification of patients with ventricular arrhythmias who are at risk for developing life-threatening ventricular arrhythmias” with Class IIa Level of Evidence A. The MMA method is FDA-cleared to measure TWA in ambulatory ECG recordings.

It is additionally relevant that since the submission of our request, the International Society of Holter and Noninvasive Electrocardiology (ISHNE) endorsed the MMA method. (See public comment by Ryszard Piotrowicz, M.D., Ph.D., President, ISHNE, on November 13, 2014).

We note that during the present comment period, world leaders in the field of sudden death risk stratification and individuals who have established the guidelines for ICD implantation have written in support of extending coverage to the MMA method. These include Dr. Arthur Moss, the Principal Investigator of the MADIT-II trial, which provided the primary LVEF guideline for ICD implantation, and a member of the 2006 AHA/ACC/ESC guidelines writing committee recommending TWA assessment to “evaluate risk or to judge therapy” and to “improve diagnosis and risk stratification”; Heikki Huikuri, Principal Investigator of major trials in SCD risk assessment and lead author of the widely referenced New England Journal of Medicine article, “Sudden Death due to Cardiac Arrhythmias” (2001;345:1473–82); Mark E. Josephson, international educator and author of the definitive textbook, Clinical Cardiac Electrophysiology; Takanori Ikeda, the TWA investigator with the largest experience with the SA method; and Marek Malik, Principal Investigator of trials on SCD risk assessment and chairman of key consensus statements in leading journals, as well as others.

We trust that based on the new information provided during this public comment period and the high level of support of extending coverage to the MMA method by leaders in the field that CMS will not continue to restrict coverage exclusively to the SA method.

Sincerely,

Ian Rowlandson
Chief Scientist for Diagnostic ECG, GE Healthcare

Less

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alterna006Es (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

I am writing to express my concern that the Spectral Analytic method is being portrayed as a monolithic, immutable gold standard.

In fact, since the favorable coverage decision for the Spectral Analytic method in 2006 (CAG-00293N), which was based on only 12 studies that enrolled 2,632 patients, there have been major nonpredictive studies with this method, including the sizeable SCD-HeFT substudy (Gold et al, Circulation 2008;118:2022–8), MASTER (Chow et al, J Am Coll Cardiol 2008;52:1607–15), CARISMA (Huikuri et al, Eur Heart J 2009;30:689–98), and the Glasgow studies (Jackson et al, European Journal of Heart Failure 2013;15:1253–1261).

Significant difficulties with the Spectral Analytic method are apparently at the root of these failures, particularly its high rate of indeterminate test results, in the range of 20-40%, which limits the test’s utility, as pointed out by Fishman et al (Circulation 2010;122;2335-2348). This difficulty has required the introduction of changing contingency rules. First of these is by Bloomfield et al (J Cardiovasc Electrophysiol 2002;13:502–12) with complex “A,” “B,” and “C” rules, which have been variably employed. More recently, Kaufman et al (J Am Coll Cardiol 2006;48:1399–404) attempted to address the indeterminacy rate problem by introducing the categories of “transient TWA episodes” of unspecified duration, ventricular ectopy, and patients’ inability to achieve a target heart rate, which occurred in 50% of cases as “nonnegative” or “indeterminate due to patient factors.” There is an inherent limitation in classifying “failure to achieve heart rate” as a “nonnegative test,” as this factor constitutes a measure of chronotropic incompetence and not a determination of MTWA. For example, the heart rate blunting effect of beta-adrenergic blocking agents can affect the reclassification of the Spectral Analytic method test, suggesting a deleterious effect, when in fact heart rate reduction is a well-established basis whereby this class of agents reduces risk for SCD. The meta-analysis by Chan et al (J Cardiovasc Electrophysiol 2010;21:1009–14) demonstrated that washing out beta-blockers in order to raise heart rate, and their subsequent resumption, reduces the predictive power of the test by a factor of four.

By comparison, the MMA method has no target heart rate requirement, thus avoiding the necessity of washing out chronic medications and resulting high indeterminate rate. All MMA studies that enrolled =100 patients, CMS’s criterion in the proposed decision memo, have predicted endpoints. The few studies that used noncommercial, “homebrew,” custom software have yielded non-predictive results, for example, Cox et al (Pacing Clin Electrophysiol. 2007;30:352-358) and Kim et al (Circ Arrhythm Electrophysiol 2014;7:259-66); they are not relevant to the current application. These ad hoc methods lack the well-tested ECG, noise, respiratory, and motion filters as well as the QRS-aligned templates, which permit overreading to verify the presence and level of MTWA and are integral to the MMA method as implemented in the commercial software.

While the signal processing methods differ, the conceptual approach is similar. SA measures MTWA at 0.5 cycle/beat while MMA assumes a dichotomous distribution of successive beats in ABABAB pattern.

Based on this evidence, I sincerely hope that CMS will reverse its decision, as MMA-based MTWA analysis is a valuable test which is at least equivalent to the SA method in identification of individuals who are at risk for SCD and is thereby capable of improving health outcomes for CMS recipients.
Sincerely,

Less

November 21, 2014

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

As co-inventor, together with Dr. Bruce Nearing, of the Modified Moving Average (MMA) method for microvolt T-wave alternans (MTWA) assessment, licensed by GE Healthcare, we greatly appreciate the careful evaluation of the request for National Coverage of this test. The key study question that guided the review, as stated in section VIII. CMS Analysis, B. Answer to Key Question, appears to be the following: Is the evidence sufficient to conclude that MTWA testing using the MMA method improves health outcomes for Medicare beneficiaries who are candidates for ICD placement? As GE Healthcare’s application is a claim of equivalence in predictive capacity to the currently covered Spectral Analytical (SA) Method (CAG-00293N), it would seem appropriate that matching criteria would be applied in the current evaluation of the MMA method.

As reviewed in detail below, the MMA method has been standardized and guidelines and recommendations published in an international MTWA Consensus Guideline statement (J Am Coll Cardiol 2011;58:1309–24) in terms of cutpoints, lead selection, and update factor. QRS-aligned templates (Fig. 1, page 1310), which are published in a majority of MMA studies, are an integral function of the software for MMA testing for both routine exercise testing and ambulatory ECG monitoring. These templates enable the cardiologist to inspect and verify the MTWA level as a distinct separation in the J-T interval in successive A and B beats in the ABABAB beat stream, the hallmark of the MTWA phenomenon. Verification can be made down to 20 microvolts, well below the 47-microvolt cut point of abnormality using the MMA method. Thus nonalternating patterns due to noise and respiration can readily be rejected by individuals analyzing the automated display. By contrast, the SA method, which performs the analysis in the frequency domain, does not provide means for the cardiologist to verify the presence of MTWA, although it is clearly visible in MMA templates in patients with comparable disease states including CAD, heart failure, and ischemic and nonischemic cardiomyopathies.

The authors of the MTWA Consensus Guideline concluded that the MMA studies as a whole have demonstrated equivalent prediction to SA studies in diverse populations at risk for SCD with hazard ratios for heightened risk for SCD of 4.4 to 22.6 compared to 2.1 to 16.8 for the SA method. A head-to-head comparison of MMA with the SA method has also been published (Exner et al J Am Coll Cardiol 2007;50:2275–84), showing strong agreement based on formal statistical comparisons. This important demonstration is consistent with the fact that the SA and MMA methods measure the same electrophysiologic entity, namely microvolt levels of TWA, with a remarkably high linear correlation coefficient of 0.95 with identical signal inputs (Hostetler et al Comput Cardiol 2005;483-6). As presented in the detail below, sizeable populations have been studied with disease status and age range comparable to U.S. Medicare beneficiaries. Use of control subjects from within the studies and the negative predictive values for the two methods are comparable.

As is the case for the SA as well as the MMA method, no prospective randomized controlled trial for MTWA-guided therapy has been performed, because meeting this requirement would necessitate withholding ICDs from patients who meet well-established MADIT-II criteria. Furthermore, ICD firing provides an unreliable surrogate for SCD (Ellenbogen et al Circulation 2006;113:776–82), and the devices themselves are proarrhythmic (Germano et al Am J Cardiol 2006;97:1255–1261). These factors are thought to explain in part the lack of prediction in several MTWA trials using the SA method (Hohnloser et al Heart Rhythm 2009;6:S36–44). SCD remains the preferred endpoint in studying primary prevention patients (Fishman et al Circulation 2010;122;2335-2348).

A major advantage that the MMA method offers to CMS beneficiaries is circumvention of the target-heart rate requirement of 105-110 beats/min during exercise by the SA method. This rate requirement is problematic as evidenced by the fact that in half of recent studies with the SA method it has been deemed necessary to washout beta-blockers in order to achieve the required heart rate. This protocol resulted in a 4-fold decrease in prediction, with an average relative risk of only 1.40 (95% CI: 1.06-1.84) (Chan et al J Cardiovasc Electrophysiol 2010;21:1009-1014). Even more importantly, as pointed out by these authors, because beta-blockers decrease TWA consistent with their reduction in SCD (Olsson et al Eur Heart J 1992;13:28 –32), screening patients whose beta-blockade therapy has been washed out, with resumption after the test, carries the risk of introducing false positive results. This practice could lead to incorrect classification of these patients as being “at risk” with potential inappropriate candidacy for ICD implantation. As nearly one third of CMS beneficiaries take beta-blockers (Psaty et al Arch Internal Med 2002;162:2325-2332), this limitation of the SA test is significant. By contrast, the MMA method has no heart rate requirement and can be performed from ambulatory ECG monitoring and routine, symptom-limited exercise stress testing recordings. The former platform also confers the benefit of MTWA testing in CMS beneficiaries who are unable to perform an exercise test.

The facts and the detailed clarifications below, as well as the public comments by world leaders in SCD risk stratification and ICD implantation support the claim that MMA is at least equivalently predictive to the SA method, which was previously deemed to improve health outcomes for Medicare beneficiaries who are candidates for ICD placement, the current study question. The MMA method has been endorsed by the International Society of Holter and Noninvasive Electrocardiology (ISHNE). (See public comment by Ryszard Piotrowicz, M.D., Ph.D., President, ISHNE, on November 13, 2014).

With restriction of coverage to the SA method, CMS beneficiaries who cannot perform a target-heart rate exercise test due to advanced age, disease, and/or chronic medications such as the widely employed beta-blockers are denied MTWA testing for sudden death risk assessment.

Hopefully, CMS will reassess the merit of MMA and revise its proposed noncoverage decision, as MMA analysis of MTWA is particularly well-suited for use in CMS beneficiaries as an accurate, patient-friendly method that can be employed in the routine flow of clinical evaluation during either a standard, symptom-limited exercise test or ambulatory ECG monitoring.

Respectfully submitted,

Dr. Verrier
Associate Professor of Medicine, Harvard Medical School
Beth Israel Deaconess Medical Center

POINT #1: MTWA GUIDELINE STATEMENTS ARE PUBLISHED IN MAJOR JOURNALS

November 17, 2014

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

As I indicated in my prior comment in May 2014, the area of emphasis of my clinical and research activity is assessment of cardiovascular risk in patients with ischemic heart disease. This work is in the context both of intensive care, as I have been the Co-Director of the Samuel A. Levine Cardiac Unit at Brigham & Women’s Hospital, and in the use ambulatory ECG and exercise stress testing in evaluating risk in patients with coronary artery disease.

Given the substantial evidence that supports MMA-based TWA and its equivalence or superiority to the Spectral Analytical method, I was frankly surprised by the preliminary decision memo not to cover this valuable test. This is particularly the case as the MMA method is the only FDA-cleared technology for TWA testing during ambulatory ECG monitoring, a mainstream clinical platform. Ambulatory ECG-based TWA monitoring received a class I, level of evidence A, indication in 2006 AHA guidelines (Zipes et al 2004, DOI: 10.1161/CIRCULATIONAHA.106.178233).

I strongly urge CMS to reconsider its decision and to cover this test, which provides useful data applicable to CMS patients and the pathologic conditions that afflict them, including most notably coronary artery disease and heart failure. The availability of this test is likely to result in better healthcare decisions through more accurate sudden death risk assessment and thereby to improve medical management including ICD implantation.

Sincerely,

Peter H. Stone, M.D.
Professor of Medicine, Harvard Medical School
Senior Physician, Brigham & Women’s Hospital
Director, Vascular Profiling Research Group
Director, AECG Monitoring and ECG Core Laboratory

Less

21 November 2014
Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

My area of expertise is signal processing of electrocardiograms with respect to sudden death risk stratification. I have published original articles on the topic of T-wave alternans and have reviewed the underlying principles in our article, “Methodological principles of T wave alternans analysis: a unified framework (IEEE Trans Biomed Eng 2005;52:599–613),” which was cited in the CMS proposed decision memo CAG-000293R2. In addition, I was a coauthor of the TWA Consensus Guideline, which was published in JACC (2011;58:1309–24).

Therefore, I read the proposed decision memo with interest but was surprised by CMS’ negative determination regarding coverage of the TWA analysis by the MMA method. This opinion does not reflect the substantial literature and fundamental principles that this approach employs to detect TWA levels accurately. The robustness of the MMA technique with respect to noise rejection and accurate quantification was discussed in detail in our 2005 review. In the TWA Consensus Guideline, the appropriate use of the MMA method was clearly spelled out in detail, including cutpoints, update factor, and lead selection, etc. Also, the international panel of experts on microvolt TWA concluded that the MMA and Spectral methods were similar in terms of prediction based both on the head-to-head study by Exner et al (2007) and on the populations studied as a whole.

From a technical point of view, it has been shown that other signal processing methods apart from the Spectral Method allow a similar or improved quantification of the TWA level. The MMA method has sound clinical evidence supporting its use and employs a robust signal-processing approach. Therefore, based on my research experience, I believe that the incorporation to clinical routine of TWA tests based on solid and well-assessed methodologies (as it is the MMA method) should be supported and not excluded from reimbursement, as it has been shown to be a valuable tool in assessing risk for life-threatening ventricular arrhythmias.

Sincerely,

Juan Pablo Martinez, Ph.D.
Biomedical Signal Processing and Computational Simulation Group
Aragon Institute of Engineering Research
University of Zaragoza, Zaragoza, Spain

Less

At the very close of the initial public comment period, Dr. Richard Cohen, inventor of the Spectral Analytic method of T-wave alternans analysis, and Mr. Roderick de Greef, the President and CEO of Cambridge Cardiac Technologies, Inc., current producer of the spectral technology, argued that the spectral method is superior to the MMA method due to its unique ability to compensate for noise and artifact. I find this argument to be rather specious. In presenting printouts of all the leads and trends using the MMA method, the clinician can clearly interpret the quality of the tracings and the information provided. In fact, annotation is provided and unacceptable artifact is clearly presented and specifically indicated in the tracings. Also indicated is the noise level cut off (20 microvolts). Unlike the spectral method of reporting test results (normal, abnormal, indeterminate) the Modified Moving Average (MMA) method reports actual microvolt levels. Thus, the 47 µV cutoff falls within a continuum of cardiac electrical instability. The MMA provides the necessary granularity, unavailable with the Spectral method, that allows the clinician and patient to make informed decisions regarding treatment, prognosis and reducing the risk of sudden cardiac death.

Electrocardiographic tracings for decades provided no interpretation but computerization now provides such an interpretation for the physician's confirmation. The physician still is obligated to review the tracings, which have been processed by computer algorithms, to be certain the interpretation is accurate. This is in fact what the MMA method provides: an opportunity to review the hard data and judge if the quality warrants significance. This is unavailable with the spectral method. I cannot imagine when computerized electrocardiographic interpretations became available that they would be simply stated as "normal, abnormal, or indeterminate," as is the case for the Spectral Method. Test results such as that would clearly place the patient in great danger.

Randomized controlled trials to prove a survival benefit for MMA over the Spectral method should not be a requirement for reimbursement approval for this has never been a CMS requirement for approval of new and improved technologies. The obvious clinical enhancement of B-mode echocardiography over M-mode was enough to warrant reimbursement without undergoing survival studies. Doppler interrogation became clearly advantageous and reimbursable without patients being subjected to prospective studies proving a survival advantage in order for the superiority of this technology to be reimbursed. The list of examples is exhaustive and the MMA has at least demonstrated its equivalency to the spectral method.

Approximately 40 cardiologists in my locale have access to the spectral method of T-wave alternans analysis. None of them continues to use this modality when treating their patients at risk for sudden cardiac death. Reasons for their discontinuance include the expense of the proprietary electrodes, cumbersome patient workflow in scheduling a non-standard exercise test, and overall dissatisfaction with the reports. On the other hand, across more than 10 years of experience with MMA, I have not experienced any of these hindrances. Although good data exist for the utilization of the spectral method if, in practice, the testing is too burdensome to utilize, then the methodology may devolve into a “bookshelf” research curiosity.

I strongly urge CMS to reverse its previous decision and not to deny its subscribers the valuable and potentially life-saving benefit of the Modified Moving Average T-wave alternans testing procedure.

Less

November 17, 2014

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

As I have previously stated, I have done considerable research in ECG-based sudden cardiac death risk stratification, with a special emphasis on ambulatory ECG recordings. When I learned that CMS was reconsidering coverage for the Modified Moving Average (MMA) method, I was very pleased, in light of the published evidence supporting this method as a valuable and non-invasive risk stratification tool. I was surprised to learn that the proposed decision memo was negative despite the extensive literature supporting MMA.

In reading the proposed decision memo, I noted some inaccuracies in the summary of my own studies, and I am writing to request that the record be corrected. I first want to correct the impression on the part of the reviewer that TWA analysis in our study on acute post-myocardial infarction patients with left ventricular dysfunction from the EPHESUS (J Cardiovasc Electrophysiol 2008;19:1037-1042) was not blinded and that the cutpoints were not pre-specified. It is true that this was a matched 2:1 nested case:control study, so the cases and controls certainly had known outcomes, but once they were selected, we were totally blinded both in our Holter-based TWA analysis and in our data entry, to this information. Also, cutpoints for significant TWA were pre-specified, but we were surprised to find that maximum TWA did not occur at the time of maximum heart rate on the recording, a result that supports the potential for 24-hour Holter recordings to identify high risk people.

Also, I was surprised that it was concluded that the MMA literature is not relevant to the CMS population of older adults. In the NIH-sponsored Cardiovascular Health Study (Journal of Electrocardiology 2010;43:251–259), we analyzed Holter recordings in sample of Americans = 65 years old and found that TWA on Holter was associated with increased risk for SCD during the following ~5 years with a hazard ratio of 4.8 (95% CI: 1.48-15.81, p = 0.009). This was also a nested case:control study and again, once participants were selected, we were blinded to their status during the analysis and once again significant TWA did not usually occur at the maximum heart rate of the recording. Based on the age range of this population and the disease conditions that affect them, it can be argued that this analysis is directly relevant to Medicare enrollees.

I also want to reemphasize that MMA enables TWA assessment on ambulatory ECGs. Thus, no patients had to be excluded from the test on the basis of inability to perform an exercise test as a result of frailty and locomotor dysfunction in either of the studies that I published. From personal experience with this cohort, and in accord with the fact that older individuals may have difficulties in exercising, the availability of a test that enables TWA assessment on ambulatory ECG monitoring is a significant advantage for CMS subscribers. Furthermore, real-life MMA TWA measurement on Holter might capture times of instability that would be missed in a clinical test appointment. The MMA-based TWA test is unique in this regard, as it is FDA-cleared. TWA analysis from ambulatory ECG recordings is recommended by the American Heart Association and American College of Cardiology with Class I, level of evidence A (J Am Coll Cardiol 2006;48:e401). The recent TWA Consensus Guideline (JACC 2011;58:1309–24) also supports the use of the MMA method.

Accordingly, I continue to believe that if CMS is willing to cover the spectral method for risk stratifying patients by finding those with TWA, CMS should allow coverage of the MMA TWA test as well. It is hard to justify only supporting the alternative Spectral Analytical Method, which is less generalizable to an elderly population since it can only be performed during an exercise test and only in people who can sustain a target heart rate of 105-110 beats/min, conditions that may not be applicable to a significant number of patients in the CMS age range.

Hopefully, after considering this and other evidence provided by leaders in sudden cardiac death risk stratification who have already commented in favor of coverage of the MMA method during this period, CMS will reverse its current position in favor of supporting the MMA test as well.

Sincerely,

Less

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

I must confess that I was astonished by CMS’ preliminary decision memo not to cover the MMA method. This decision is at odds with a sizable literature of studies that enrolled >5000 patients and all studies that enrolled >100 patients predicted endpoints. The MMA method is supported by a prestigious multi-author TWA Consensus Guideline published in JACC and in a position paper of the AHA/ACC recommending Holter-base TWA, for which MMA is uniquely cleared by the FDA. I appreciate that the current CMS memo recognizes my early observation 40 years ago of the importance of TWA in risk assessment for sudden cardiac death. However, quantitative assessment with the MMA method, with the opportunity to evaluate QRS aligned templates to verify TWA level, is a major step forward.

In the first study to use this method clinically, published over a decade ago, for which I was the senior author, TWA >47 microvolts was established as a key cutpoint; this determination has withstood the test of time as it is the current cutpoint for abnormality.

This cutpoint, along with the appropriate leads to be employed, update factors, and other technical parameters are not only clearly stated in the TWA Consensus Guideline but also in the TWA Physicians Guide, which is distributed with the FDA-cleared software. As noted by several of my distinguished colleagues in arrhythmia risk stratification, including Arthur Moss, the Principal Investigator of the MADIT-II study, Heikki Huikuri, a thought leader in SCD research, and Marek Malik, an internationally recognized authority in SCD risk stratification, the MMA method constitutes a valuable diagnostic tool in our armamentarium for SCD risk stratification.

MMA predicted SCD after multivariate analysis including LVEF. The age range and diseases studied such as CAD and heart failure are highly relevant to the CMS population.

Thus, MMA goes beyond LVEF dysfunction as a risk stratifier for arrhythmia. LVEF has limited sensitivity and specificity as many patients in whom ICDs are implanted prove not to need them, while many who are denied a device based on this marker of mechanical function die suddenly.

I believe that it would be a grave mistake to deny coverage to the CMS population of the MMA method for a risk-stratification tool that offers accuracy and compatibility within the routine flow of clinical evaluation including exercise stress testing and uniquely ambulatory ECG monitoring. The latter platform is extremely important, as many patients in the CMS age group are unable to perform an exercise test due to frailty and limited physical capacity as a result of disease or advancing age.

I strongly appeal to CMS to reconsider its proposed decision memo and not to deny coverage of the MMA method, which is at least equivalent if not superior to the currently supported Spectral method.

Sincerely,

Peter J. Schwartz, MD

Less

November 11, 2014

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

This letter serves as an addition to my prior comment made in May 2014 as CMS was considering this issue. I am a clinical cardiac electrophysiologist with a primary interest in the problem of sudden cardiac death and I make decisions regarding appropriate ICD implantation as part of my routine clinical practice. In my previous comment on the National Coverage Analysis of Modified Moving Average (MMA)-based T-wave alternans (TWA) analysis (CAG-00293R2), I recommended unequivocally the use of the MMA method for TWA analysis in evaluating sudden cardiac death (SCD) risk. My support for this technology is based on the fact that there is extensive scientific and clinical evidence and published guidelines. The American Heart Association/American College of Cardiology/ European Society of Cardiology 2006 recommendation (Circulation 2006;114;e385-e484) endorses the use of ambulatory as well as exercise-based TWA for arrhythmia risk evaluation. MMA is the only methodology that is FDA-cleared for ambulatory ECG-based analysis of TWA.

The evidence in favor of coverage for MMA is derived from the study of >5000 patients. It is arguably at least equivalent to the Spectral Method in its predictive capacity, as hazard ratios are similar for the studies as a whole and specifically in the head-to-head comparison in the REFINE study by Exner et al (2007). Eleven TWA studies use SCD as an endpoint and include data on positive and negative predictive values in most; MMA and the Spectral Method are similar in terms of negative predictive value. Therefore, it is puzzling to me why the proposed decision memo does not support coverage of the MMA method.

This negative decision, in my opinion, would deprive CMS patients of a valuable, potentially life-saving test. The populations that have been tested are highly relevant to the CMS population. MMA studies were published in peer-reviewed journals that are of the highest quality in cardiac electrophysiology and constitute a valuable source of information for clinical decision-making, beyond the limitations of left ventricular ejection fraction (LVEF). Sole use of LVEF currently results in excess implantation of ICDs, as conservatively for every 8 patients who receive devices, discharge does not occur in 7. MMA-based TWA studies demonstrate predictive power in multivariate analysis beyond LVEF as well as standard clinical variables and traditional risk markers.

Therefore, I strongly urge CMS to reconsider its proposed decision and to cover the MMA method of TWA analysis.

Less

13 November 2014
Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244
Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

The Officers and Board of Trustees of the International Society for Holter and Noninvasive Electrocardiology (ISHNE) is writing to register our Society’s endorsement of Modified Moving Average (MMA) analysis of T-wave alternans for sudden cardiac death risk stratification. Our decision to endorse this method is directly based on the evidence provided in the TWA Consensus Guideline report, which was sponsored by our Society and prepared by an official writing committee comprised of Drs. Verrier, Malik, and Klingenheben, in collaboration with international experts in both the Spectral Analytic and MMA methods, including Drs. El-Sherif, Exner, Hohnloser, Ikeda, Martinez, Narayan, and Nieminen, and, sadly, the late Dr. Rosenbaum. In addition to ISHNE, the document was endorsed by three other societies, namely, the Japanese Circulation Society, the Computers in Cardiology Working Group on e-Cardiology of the European Society of Cardiology, and the European Cardiac Arrhythmia Society. The article underwent peer-review before being published in the prestigious Journal of the American College of Cardiology (2011:58:1309–24). We were pleased to see the endorsement in his public comment by our colleague, Dr. Arthur Moss, a pioneer in ambulatory ECG monitoring, who was a founder and the first President of our Society. He was also the Principal Investigator of the MADIT-II trial, which provided the main guidelines for ICD implantation, a subject that is directly relevant to the Proposed Decision Memo.

The TWA Consensus Guideline summarized knowledge to-date regarding the MMA and Spectral methods of TWA analysis, compared their clinical study results and analytical approaches, and provided specific recommendations regarding cutpoints, update factors, lead selection, and interpretation. An important conclusion was reached, namely, “Hazard ratios for prediction by the Spectral and MMA methods are similar, whether in the same population (Exner et al JACC 2007;50:2275–84) or in studies overall” (Tables 1 and 2 on pages 1312-1314). Finally, the consensus guideline recommended, “It is reasonable to consider TWA evaluation whenever there is suspicion of vulnerability to lethal cardiac arrhythmias” (page 1321).

Thus, we hope that the results of this comprehensive, unbiased expert review will be duly considered in your final decision memo.

Sincerely,

Less

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard, Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

My career has been dedicated to the advancement of knowledge regarding the mechanisms responsible for malignant cardiac arrhythmias and to medical and device-based management of these patients. It is clear to me that the current guidelines for ICD implantation based on LVEF have major shortcomings, as it is well known that many patients who are implanted with these devices based on MADIT-II criteria do not experience arrhythmias requiring ICD shock. SCD is an electrophysiologic event and it is not surprising that a parameter based on mechanical function, LVEF, has such significant limitations.

I find ECG-based assessment an inherently more logical approach. T-wave alternans (TWA) assessment by the Spectral Analytic method has been shown useful in a number of studies. However, in investigations in which ICDs had been implanted, prediction has been limited. At this point, it is plausible that the failure of this method to predict outcomes in the ICD studies (MASTER, SCD-HeFT) may be attributable at least in part to the use of ICD discharge as a surrogate endpoint. In addition, the devices themselves can be proarrhythmic, as my colleagues and I have recently reported (Germano et al Am J Cardiol 2008;97:1255-1261). I do not believe that the approach of washing out medications to allow the patient to reach a target heart rate of 105-110 beats/min, with resumption of medications following the test, as was done in these studies, is sound.

The MMA method, which does not have a heart-rate requirement, circumvents this significant conceptual and practical limitation. As far as I am aware in reading the peer-reviewed MMA literature, all studies using this method have predicted outcomes. This fact is clearly stated in the TWA Consensus Guideline (JACC 2011;58:1309–24), which was authored by international experts in both the Spectral and MMA methods and was sponsored by 4 international societies.

In my opinion, based on the clinical evidence provided, on the fact that the MMA method has received FDA-regulatory clearance for SCD risk assessment, and on the cost-efficiency of the test, it is my opinion that it would be inappropriate to exclude this method selectively from CMS reimbursement, as it is an equally valid test as the Spectral Method and is more user friendly for patients and their cardiologists.

Therefore, I recommend deletion of the words “only when the Spectral Analytic method is used” from the decision.

Sincerely,

Less

The primary focus of my clinical and research activities over the course of many years has been to address the risk for life-threatening arrhythmias in my patients individually and in general regarding those who suffer from coronary artery disease and the progression of heart failure.

As a cardiologist dealing with malignant arrhythmias and Chief of the Main Line Health Division of Cardiovascular Diseases, which includes an extensive healthcare network in the greater Philadelphia area, I deal daily with management of patients who are at risk for sudden cardiac death (SCD) including making decisions regarding ICD implantation. It is clear that while the MADIT-II criterion of LVEF =35% is valuable, it has important limitations as it can lead to a 7-fold excess of implantations. Furthermore, many patients who are at risk have relatively preserved LVEF, >35%.

Thus, there is a great need for improved risk assessment beyond LVEF. Microvolt T-wave alternans (TWA), both on scientific and clinical grounds provides important additional information regarding SCD risk. While the spectral method has been shown promising in this regard, it is nevertheless an imperfect test, as it yields indeterminate results in up to 40% of patients. The approach of using failure to achieve and maintain a target heart rate of 105-110 beats/min to indicate a positive test is particularly problematic in the CMS population, who have frailties due to aging and disease affecting their locomotor capabilities. Furthermore, the heart rate requirement constitutes a test of chronotropic competence and is not a measure of cardiac electrical instability, as is the case for the genuine TWA endpoint. The alternative option is to wash out chronic medications for the test with resumption following it; this approach is the likely basis for the failure of the SCD-HeFT and MASTER TWA studies. In some of the hospitals in our system, the Spectral Analytic Method testing has been employed from time to time. The feedback from our cardiologists has been that the requirements for a non-standardized exercise test with separate scheduling and the use of the oversized specialized electrodes, which can cause welts, are problematic.

By contrast, the more user-friendly Modified Moving Average (MMA) method can be performed in the course of routine clinical activity, including symptom-limited exercise testing with standard electrodes and ambulatory ECG monitoring. A sound and substantial clinical literature supports this test, and clear guidelines for its use are provided in the TWA Consensus Guideline in JACC (2011;58:1309–24) as well as in the manufacturers’ TWA Physicians Guide, which is included in the MMA software package.

Accordingly, I do not believe that it is correct to deny coverage of this valuable test, which is at least equivalent or superior to the currently reimbursed Spectral Analytic method. I strongly urge CMS to reverse its Proposed Decision memo.

Sincerely,

Less

Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244

9th November 2014

Dear Drs. Syrek-Jensen, Chin, Rollins and Hutter, Dear Mrs. Long:

I have had a longstanding interest in ECG-based risk stratifiers for sudden cardiac death and have, among others, chaired the writing committee of the recent Consensus Guideline on Microvolt TWA (J Am Coll Cardiol 2011; 58:1309-1324). I have therefore read the recent proposed decision memo on reimbursement of TWA testing with the Modified Moving Average method with interest.

In reading the proposed decision memo, one is left with an incorrect impression that the TWA Consensus Guideline was solely devoted to the Spectral Method, as the conclusion is drawn that there were no guidelines that specifically focused on MMA or compared MMA to SA (Section VIII Evidence, 5. Evidence-based guidelines). On the contrary, the TWA Consensus Guideline was written by an international group of experts with extensive experience and publications using both techniques. The authors were charged with the task of providing a balanced discussion of both methods and guidelines for their appropriate use. The document clearly defined the dichotomy limits, lead selection, and update factor for the MMA as well as the Spectral Method. Moreover, it provided the head-to-head comparison of risk assessment for SCD and cardiovascular mortality showing equivalence of the two methods with specific statements of relative risks and negative predictive value. Finally, it recommended, “it is reasonable to consider TWA evaluation whenever there is suspicion of vulnerability to lethal cardiac arrhythmias” (page 1321). This statement applies equally to the MMA method and to the Spectral Method.

To date, including the period since the publication of the TWA Consensus Guideline, a sizeable number of patients have been enrolled into full-cohort TWA studies using both the Spectral and MMA TWA methods. The total enrolment for the Spectral Method studies with =100 patients is currently 11,230; 8085 patients were enrolled in studies that predicted outcomes and 3,145 (28% of total) patients were enrolled in studies that were not predictive. The total enrolment for MMA method is currently 5,230; all studies enrolling =100 patients predicted outcomes. In the studies enrolling =100 patients reviewed in CMS’ favourable decision on reimbursement of Spectral Method testing (CAG 00293N), the total enrolment was 2,453 patients, including 343 (14% of total) in the study by Grimm et al (Circulation 2003; 108:2883-2891), in which TWA did not predict outcomes. Accordingly, the current MMA-based TWA enrolment exceeds by a factor of 2 the enrolment in Spectral Method studies at the time of CMS’ favourable decision on enrolment.

Thus, one can only conclude that the position reached in the TWA Consensus Guideline that the methods are at least equivalent is still valid at present. If of interest, I shall be happy to provide you the table of the statistics that I have just described.

I also had opportunities to evaluate the TWA literature in several invited reviews. It seems obvious that the target heart-rate requirement of the Spectral Method is an important source of indeterminate test results and reduction of predictive power, especially when chronic medications such as beta-blockers are withdrawn to perform the test, with subsequent resumption (see also Chan et al, J Cardiovasc Electrocardiol 2010;21:1009-1014). The MMA method does not require a target heart rate, and indeterminate test results are limited to technical difficulties, which occur infrequently (around 3% of patients). Considered the fact that CMS patients are generally older than the overall population, with more frequent locomotors dysfunction, poor limb perfusion, and other disabilities and/or comorbidities, the utility of MMA-based TWA during ambulatory ECG monitoring is a distinct advantage. As we pointed out in the TWA Consensus Guideline, the 2006 guideline of the American Heart Association, American College of Cardiology, European Society of Cardiology, and/or Heart Rhythm Society (J Am Coll Cardiol 2006; 48:e247–346), TWA on ambulatory ECG was recommended with Class I, Level of Evidence: A “to evaluate risk or to judge therapy.” MMA is also the only FDA-approved method available for use with ambulatory ECG.

For all these reasons and based on many years of research work in the field of noninvasive sudden death risk stratification, I believe that MMA-based TWA technique is a valuable tool that should be incorporated into routine clinical practice with respect to exercise and ambulatory ECG monitoring and its use should be supported by CMS, as ultimately through improved risk assessment. The performance of MMA in terms of risk stratification is at least equal to the Spectral Method and deserves equal support. I do not understand what reason might lead to reimbursing Spectral Method tests but not MMA tests. Such a distinction is contradicts the available evidence.

I would be most grateful if you would consider these points.

Yours Sincerely,

Marek Malik

Marek Malik,
PhD, MD, DSc, DSc(Med), FACC, FESC, FHRS,
Professor of Cardiac Electrophysiology,
Senior Research Investigator,
National Heart and Lung Institute,
Imperial College of Science, Technology and Medicine
London, England

Less

5 November 2014
Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

The main focus of my clinical practice and research has been the problem of Sudden Cardiac Death (SCD) with particular emphasis on guidelines for use of ICDs. Our article, which provided the clinical evidence for use of these life-saving devices, the MADIT-II trial (N Engl J Med 2002; 346;877-883), for which I was Principal Investigator, demonstrated the value of depressed left ventricular function (LVEF) in post-myocardial infarction patients and has been widely employed to guide physicians worldwide in their decision making regarding ICD implantation. While I believe that LVEF is a critical index of risk for SCD, other noninvasive ECG parameters need to be considered.

I have followed the literature on T-wave alternans (TWA) for many years and have published articles on this subject. I find the approach of quantifying TWA with advanced analytical techniques such as the Spectral Analytic and Modified Moving Average (MMA) methods inherently attractive. In my opinion, it is worthwhile for CMS to cover use of these techniques. However, I do not agree that covering one and not the other of these two FDA-cleared methods is appropriate.

When I read the Preliminary Decision Memo and its decision not to cover the MMA method, I was frankly surprised, given the high quality of the literature supporting the method’s utility. The TWA Consensus Guideline (J Am Coll Cardiol 2011; 58:1309–24) was written by an international panel of experts in both methods and published in Journal of the American College of Cardiology under the sponsorship of The International Society for Holter and Noninvasive Electrocardiology, Japanese Circulation Society, the Working Group on e-Cardiology of the European Society of Cardiology, and the European Cardiac Arrhythmia Society. This document clearly shows the equivalence of the two methods.

In addition, the MMA method offers, in my view, an extremely important feature, namely, the possibility of making the TWA measurement from ambulatory ECG recordings. This platform offers flexibility and ease of use for patients, particularly the elderly. It is in part for this reason as well as its value in risk stratification that the 2006 AHA/ACC guideline authors, including myself, granted a Class I, Level of Recommendation A, classification (Circulation 2006;114;e385-e484) to TWA monitoring with ambulatory ECG.

I find that the literature on MMA-based TWA analysis highly relevant to the CMS population. The study groups encompass the same underlying pathology, coronary artery disease, heart failure, and ischemic and nonischemic cardiomyopathy, as those that are prevalent in the CMS population. It is also germane that the MMA studies included patients in the “grey zone,” with moderately depressed LVEF, as this cohort constitutes a large number of patients at risk for SCD, in whom LVEF is less reliable as an SCD indicator.

A noteworthy aspect of the MMA approach is that it incorporates quantitative analysis, inasmuch as the actual TWA level is recorded, rather than employing a single cutpoint or a binary classification as solely “positive” or “negative.” By providing an actual value in microvolts, the level of risk across time can be tracked in response to major events such as myocardial infarction and post-myocardial remodeling and to medical interventions. I am impressed by the recent study by Kentta et al (Am J Cardiol 2014;114:832-837) in collaboration with Dr. Huikuri, a leader in the field of SCD risk stratification, and its demonstration that MMA-based TWA reflects the benefit of exercise rehabilitation in patients with coronary artery disease with and without diabetes. Should these findings be confirmed in subsequent studies, particularly with outcomes, the health benefit to the CMS population could be substantial. Supporting the use of this tool among practicing cardiologists should be of interest to CMS as part of its commitment to exercise rehabilitation, which has been amply shown to reduce risk for myocardial infarction and cardiac mortality (Hammill et al, Circulation 2010;121:63-70).

In summary, I firmly believe that the MMA method for TWA measurement is based on sound scientific and clinical evidence. Reimbursement for use of this method for the evaluation of CMS patients should not be denied.

Sincerely,

Less

I would like to have comments to a recent decision by Center for Medicare and Medicaid Services regarding the application for reimbursement of the Modified Moving Average (MMA) analysis of TWA.
First of all, I appreciate that you reviewed in detail all of the articles that our applicants cited in support of the application. On the review process, I have heard that a particular question i.e., “whether or not the individual who analyzed TWA levels was blinded regarding clinical outcomes”, was raised regarding our study with Dr. Sakaki and coworkers published in Heart Rhythm in 2009. Since I was a supervisor of their study, I would like to answer to the question as follows. After recordings of 24-hour Holter ECGs, investigators (Dr Sakaki and coworkers) immediately analyzed the MMA-TWA with manual editions, and then our staffs followed the patients. I believe that this protocol means that "individual who analyzed TWA levels was blinded regarding clinical outcomes". Thus, there was clearly no opportunity for bias in declaring the TWA values. In addition, I wonder regarding why the findings of the TWA Consensus Guideline (J Am Coll Cardiol 2011) were ignored. I do hope that Center for Medicare and Medicaid Services will reconsider its decision and reimburse TWA testing by the MMA method.

Less

4 November 2014
Tamara Syrek-Jensen, J.D., Director, Coverage and Analysis Group
Joseph Chin, M.D., M.S., Acting Deputy Director, Coverage and Analysis Group
James Rollins, M.D., Ph.D., Division Director
Kimberly Long, Lead Analyst
Joseph Hutter, M.D., M.A., Medical Officer
Centers for Medicare and Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244

Re: Proposed Decision Memorandum: Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Chin, Rollins, and Hutter and Mrs. Long:

My area of expertise for more than three decades has been sudden death risk stratification. Together, Drs. Myerburg, Castellanos, and I worked to define the relative importance of sudden death incidence and prevalence in a widely cited article in the New England Journal of Medicine (2001;345:1473-1482). Over the years, together with international colleagues, we have explored a variety of noninvasive risk stratifiers and compared their merit to the main contemporary standard, left ventricular ejection fraction (LVEF). It is well known that this parameter, although widely used in decision making with respect to implantation of cardioverter-defibrillators (ICDs), has many limitations, as in many individuals who receive a device, it does not discharge, and a larger number of individuals with preserved LVEF succumb to sudden death.

Our group has direct experience with T-wave alternans (TWA). While the Spectral Method has been extensively used, there have been a number of studies in which this analytical approach has proved negative, including two of our studies, namely, Tapanainen et al (J Cardiovasc Electrophysiol 2001;12:645-652) and the CARISMA study (European Heart Journal 2009;30:689–698). While it is not entirely clear why these studies did not demonstrate prediction by Spectral Method TWA, a plausible basis is the target heart rate requirement of 105 to 110 beats/min. In the prior study, we found that sustained TWA during the pre-discharge TWA exercise test after acute myocardial infarction does not indicate increased risk for mortality. In the multinational CARISMA study, in 312 post-myocardial infarction patients in whom we implanted loop-recorders to document arrhythmias, we encountered similar problems with respect to achieving the target heart rate. In fact, in a number of patients, it was necessary to pace the heart during programmed electrophysiologic testing to achieve the necessary heart rate. The alternative of washout out of beta-blockers is not a satisfactory option, as these agents are known to suppress TWA and to decrease sudden death risk. Chan et al (J Cardiovasc Electrophysiol 2010;21:1009–14) reported that washout of beta-blockers during the test with subsequent resumption reduces the predictive capacity of the test by 4-fold.

By comparison, my experience with the Modified Moving Average (MMA) test has been more favorable. In the initial publication on the FINCAVAS study, which was published in the prestigious European Heart Journal (2007;28:2332–2337), my colleagues and I studied 1037 patients referred for routine exercise testing. The main indications for the exercise test were diagnosis of coronary heart disease (CHD, frequency 46%), testing vulnerability to arrhythmia during exercise (18%), and evaluation of work capacity (19%) and adequacy of the CHD treatment (24%), as well as obtaining an exercise test profile prior to an invasive operation (13%) or after an MI (10%). The relative risk of TWA for SCD was 7.4 (95% CI, 2.8–19.4; P < 0.001) and for cardiovascular mortality was 6.0 (95% CI, 2.8–12.8; P< 0.001). The negative predictive value for SCD was 99%. Although the patients were relatively young, age 58±13 years, these findings are relevant to the CMS population, as the main underlying pathology, namely, cardiac disease, is a common factor that places them at risk for SCD and is essentially identical to that of individuals =65 years of age.

More recently, my colleagues and I (Kentta et al, Am J Cardiol 2014:114:832-837) demonstrated in patients from the ARTEMIS study that exercise rehabilitation exerts significant protective effects in patients with angiographically confirmed CAD with and without diabetes. Compared to the control group of patients who did not exercise, there was a significant improvement in cardiac electrical stability as measured by MMA-based TWA. Specifically, exercise rehabilitation converted 30% of TWA tests from positive to negative in patients with CAD alone and converted 50% of TWA tests in patients with CAD plus diabetes mellitus. Should these results be confirmed in further studies, particularly after outcomes are available, the study would carry major implications as TWA could be used to track improvement in cardiac risk by exercise rehabilitation. It is also worthy to note that the patients were receiving optimal antiarrhythmic medications.

In light of CMS’ interest in cardiac rehabilitation and the convincing evidence that it reduces cardiac events (Hammill et al Circulation 2010;121:63-70), it would seem that this application of TWA to track this benefit and thereby provide incentives for compliance with rehabilitation—would be advantageous.

I wish respectfully to clarify three points in the proposed decision memo. First, the FINCAVAS study was conducted in 1037 consecutive patients, not 500, and remains one of the largest TWA studies. Second, patients were not selected with respect to the capacity to raise heart rates to >125 beats/min, which of course would carry the implications that they were extremely healthy and that the findings would not be relevant to the CMS population. In fact, peak TWA values were determined at heart rates < 125 beats/min. Furthermore, many of these patients were found to have a significant degree of cardiac disease. The mean age of the subjects was 58+13 years of age, indicating that a sizable number of patients were 65 years or older, in the CMS age range. Thus, the interpretation should be revised that the study is highly selective in regard to a healthy young patients. Third, although the claim is made that the TWA analysis was not performed in a blinded fashion, this is not the fact. The article explicitly states: “Continuous ECGs were digitally recorded at 500 Hz with CardioSoft exercise ECG system (Version 4.14, GE Healthcare, Freiburg, Germany) and analysed fully automatically by the GE Healthcare-released version of the MMA method.”

On a final note, I would add that in practical implementation of routine clinical practice, the MMA method offers attractive features. Specifically, the measurement is made during routine, symptom-limited exercise testing and does not require special scheduling and the additional expense of running the nonstandard Spectral Method test. The fact that MMA can be analyzed from ambulatory ECG records, as was the case in our cardiac rehabilitation study, is also desirable as this platform is widely used. Many patients in the Medicare population cannot perform an exercise test. We found in that 56 (17%) of the 323 consecutive patients in the study by Tapanainen et al 2001 could not perform the test due to poor overall health, locomotor abnormalities, recent or past disabling stroke, poor limb perfusion, pulmonary disease, and heart failure at the time of the test. As our patients in this study were in the Medicaid age-range (62.4±10.1 years), these limitations would be expected to apply to the US population.

Overall, I believe that it would be a major mistake to preclude reimbursement of MMA-based TWA, since there is extensive evidence that it is equivalent or superior to the Spectral Analysis method and offers clinically important advantages in its implementation.

Less

27 October 2014

Kimberly Long, Lead Analyst (kimberly.long@cms.hhs.gov)
Elizabeth Koller, M.D., Lead Medical Officer (Elizabeth.Koller@cms.hhs.gov)
Joseph Hutter, M.D., M.A., Lead Medical Officer (Joseph.Hutter@cms.hhs.gov)
Tamara Syrek-Jensen, J.D., Acting Director, Coverage and Analysis Group
Centers for Medicare and Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244

Re: National Coverage Analysis (NCA) for Microvolt T-wave Alternans (CAG-00293R2)

Dear Drs. Syrek-Jensen, Koller, and Hutter and Mrs. Long:

On May 23, the final day for submission of public comments, Dr. Richard J. Cohen, the inventor of the Spectral Method; Mr. Roderick De Greef, the CEO and President of Cambridge Cardiac Technologies, which acquired Cambridge Heart, Inc., and Dr. Otto Costantini, an investigator who published clinical studies on T-wave alternans (TWA) exclusively using the Spectral Method, submitted their remarks on the Modified Moving Average (MMA) method. As the comments were made just prior to closing the website, the following response is now provided to clarify misconceptions and factual errors, which unfortunately were reflected in CMS’ proposed decision memo of October 23, 2014.

Four main assertions were made, which contain major factual inaccuracies, which will be addressed each in turn.

Misconception #1: The MMA method is not capable of accurately measuring microvolt levels of TWA. According to the standard definition, TWA is a repeating ABAB pattern in the morphology and amplitude of the ST-segment or T wave. Not only has the Modified Moving Average (MMA) method been cleared by the FDA as having 1-microvolt precision in TWA measurement (FDA document K032513), equivalent to the Spectral Method, but the technique also provides computer-aided means for visualization of the microvolt TWA (MTWA) pattern, as it displays high-resolution templates of QRS-aligned superimposed beats both for ambulatory and exercise stress testing ECGs, which enable physicians to overread the computer-generated MTWA values.

With aid of these templates, clinical inspection of TWA in the ECG waveform is feasible down to a level of 20 µV, which is well-below the cutpoint of abnormality, namely, =47 µV. The ABAB pattern is readily distinguished from noise, which when present is evident as signal fluctuations across the entire PQRST complex. This process of overreading is recommended in the TWA Consensus Guideline (JACC 2011; 58:1309–24), in GE Healthcare’s TWA Physicians Guide, and practiced in the publications cited in the current NCD application. The availability of templates in the MMA approach offers a distinct advantage over the Spectral Method, which does not permit visualization of the ECG. TWA templates are published in these studies: Stein et al J Cardiovasc Electrophysiol 2008;19:1037-1042; Minkkinen et al J Cardiovasc Electrophysiol 2009;20:408-415; Sakaki et al Heart Rhythm 2009;6:332-337; Stein et al J Electrocardiol 2010;43:251-259; Secemsky et al Heart Rhythm 2011;8:592-598; Ren et al Cardiovascular Diabetology 2012;11:104; Hou et al J Electrocardiol 2012;45:60-65; Shimada et al Am J Cardiol 2012;110:1446-1451. Uchimara et al Am J Cardiol 2014; 25:1021-1027; Schomer et al Epilepsia, in press; Kentta et al Am J Cardiol 2014;114:832-837.

Drs. Cohen and Costantini’s and Mr. DeGreef’s contention that the MMA method is heavily contaminated by noise is also not tenable on the grounds that noise is a highly random process and would not yield any level of prediction, much less the high odds ratios shown in the MMA-based TWA literature cited. Indeed, many thousands of lines of code have been built into the commercial MMA software to permit powerful noise rejection and filtering to remove baseline wander and muscle and respiratory artifacts. The publications on this aspect of engineering cited against the MMA method are comprised mostly of abstracts that appeared several years ago and have not been published as full-length articles in the peer-reviewed literature. The only full-length articles that challenged the MMA method on this issue used custom not commercial software and thus did not make use of the commercial noise and baseline wander filters and QRS-aligned templates that are employed in the prospective clinical studies used to support the National Coverage Application.

Misconception #2: There are no standardized guidelines for use of MMA-based TWA.
The first clinical study using MMA-based TWA analysis employed a priori the 75th percentile of TWA values in patients without events as the cutpoint of abnormality. This level, =47 µV, proved to be reliable and is currently employed as well as the use of templates of QRS-aligned superimposed beats. In this first study, we employed the sensitive 1/8 update factor, which is the fraction of morphology change that an incoming beat can contribute. This recommended 1/8 update factor, as stated in the first publication in 2003, as typically used, and as restated in the TWA Consensus Guideline, provides greater sensitivity and capacity to detect transient but clinically important surges in TWA than update factors of 1/16 or 1/32. While there have been a few studies that departed from this practice, the bulk of the MMA literature followed these established cutpoints, update factor, and template overreading procedures.

Furthermore, TWA has been shown to reflect a continuum of risk. For each 20-µV increase in TWA, there is a 55% and 58% increase in risk for cardiovascular mortality and SCD, respectively. For this reason, the cutpoints of 20 and 60 microvolts have also been investigated. These guidelines established in 2003 and 2009 and confirmed in the TWA Consensus Guideline (JACC 2011; 58:1309–24) are that TWA =47 µV is considered abnormal and =60 µV is severely abnormal. The 20-µV cutpoint indicates relative absence of risk with an odds ratio < 2 and negative predictive value of 98%.

As noted in the TWA Consensus Guideline, quantification of TWA offers major advantages over the Spectral Method’s assessment of test results as “negative,” “non-negative,” or “indeterminate.” Quantification is further consistent with many established clinical measures such as arterial blood pressure, plasma lipid levels, left ventricular ejection fraction, and others, in which it has been shown useful to know individual patients’ values within a range in order to evaluate urgency of intervention and effectiveness of therapy. A comparison of quantified risk across diverse disease states is illustrated in Verrier and Ikeda Progress in Cardiovasc Dis 2013; 56:172–185, figure #5.

It is also worth noting that Royston et al (Statistics in Medicine 2006;25:127–41) have shown that treating a continuous variable, in this case TWA, as a binary function reduces its predictive power by one third to one half. For this reason and because of the fact that quantification is essential to gauge the effects of medical therapy, the authors of the TWA consensus guideline urged reporting actual TWA values when performing a TWA test, independent of method.

Misconception #3: There is no substantial literature to support the utility of MMA-based TWA in assessing risk for sudden cardiac death.
Inaccurate statements were made that no new clinical studies have been published in the medical literature since the last request for consideration in 2007. Sizeable prospective studies such as those of Sakaki and coworkers (2009) and Hou et al (2013) were summarily and falsely dismissed as not based on pre-hoc cutpoints or prospective design. In fact, these studies were peer reviewed and published in high-quality journals, enrolling 295 and 219 patients, respectively. A host of other studies also followed appropriate epidemiologic principles in terms of study design using both ambulatory-ECG and exercise based TWA. With respect to the latter, the FINCAVAS database, which enrolled nearly 3600 patients, represents the largest study in the field of TWA analysis by a factor of three. The initial study explored several cutpoints and established risk as a function of TWA levels. Analyses were performed by individuals blinded to clinical status and outcome.

Misconception #4: MMA-based TWA is not sufficiently sensitive, such that patients with low TWA levels could still be at risk and that the test would not improve health outcomes for Medicare beneficiaries. This assertion by Dr. Cohen and Mr. DeGreef is not correct that individuals who have a negative MMA-based TWA test could be placed at risk, as it has been shown that its negative predictive value for SCD is 99% (Nieminen et al 2007, Hou et al 2012). This level of performance compares more than favorably to that of the Spectral Method in the ABCD trial (Costantini et al 2009), in which the NPV at one year was 95%, and to the studies reviewed by Gehi et al (2005), who found Spectral Method’s NPV to be 97.2%.

The overall predictive strength of MMA-based TWA testing has proven to be comparable to the Spectral Method across diverse populations at risk for SCD with hazard ratios for heightened risk for SCD of 4.4 to 22.6 compared to 2.1 to 16.8 for Spectral Method (TWA Consensus Guideline JACC 2011;58:1309–24). The FDA approved use of MMA by competent healthcare professionals for risk stratification and prediction of arrhythmic death (FDA document K032513) in conjunction with other clinical information. MMA has received multi-society support. The validated algorithms have been employed since 2003 in studies enrolling >5,000 patients, which compares well to the >7,200 patients enrolled in the positive studies of the Spectral Method published since 1994. No negative MMA-based TWA studies that used the correct, established cutpoints and procedures have been published. Standards for appropriate measurement are clearly stated in the TWA Consensus Guideline.

Furthermore, MMA-based TWA testing offers inherent advantages over the Spectral Method as it allows visualization of the ECG, providing the physician with an extra dimension of confidence in the results. It does not require specialized electrodes or a nonstandard exercise stress test, allowing its use within the flow of routine clinical evaluation, facilitating implementation, and reducing costs. The MMA method offers the additional advantage that it can be used to analyze ambulatory ECGs, an important advance as many Medicare recipients are unable to perform an exercise test.

Finally, the publications offered by GE Healthcare in its application for reimbursement of TWA testing by the MMA method as well as the new publications cited below amply demonstrate that TWA analyzed by the MMA method is at least equivalent to the Spectral Method in stratifying risk for arrhythmic death. Moreover, MMA-based TWA allows physicians to track the effects of medications (Verrier and Nieminen 2010, Verrier and Malik, published online at http://dx.doi.org/10.1016/j.tcm.2014.10.006) and other therapeutic interventions as it quantifies the reduction in risk of sudden death in patients with epilepsy (Annegers et al 2000) by vagus nerve stimulation (Schomer et al, in press) and the well-established capacity of cardiac exercise rehabilitation (Hammill et al 2010) to reduce sudden death risk (Kentta et al 2014). Please see images in an attached pdf.

Below is a point-by-point response to the inaccurate claims made by Drs. Cohen and Costantini and Mr. DeGreef

1. COMMENT: Prospective studies supporting MMA-based TWA are absent. Case-control studies are not prospective.
RESPONSE: These statements are not accurate. In MMA studies, TWA levels were declared before the analysis of prediction was performed by investigators blinded to clinical status and events, and established a priori cutpoints were applied to identify patients with abnormal and severely abnormal levels of risk. In sizeable ambulatory-ECG based studies (Exner et al 2007, Sakaki et al 2009, Hou et al 2013) and in the FINCAVAS exercise-based studies (Nieminen et al 2006, 2007; Minkkinen et al 2009, Leino et al 2009, 2011), patients were enrolled into investigations predefined for TWA-based analysis of cardiovascular mortality risk including sudden cardiac death.

Hennekens and Buring (1987) stated, “In general… the distinction between a retrospective and prospective case-control study is not made.” They indicated that case-control methodology does not diminish predictive power and is an efficient approach for large databases. In case-control studies of stored data from the ATRAMI study and EPHESUS trial, investigators blinded to clinical status and outcomes prospectively analyzed the capacity of MMA-based TWA to predict SCD and cardiovascular mortality based on the a priori cutpoint of =47 µV.

2. COMMENT: MMA is not measuring microvolt TWA but noise.
RESPONSE: This statement is not accurate. MMA software (Nearing and Verrier 2002) reports peak TWA levels as the test result and provides templates of superimposed beats showing the ABAB pattern characteristic of TWA to allow physicians to review and confirm results. TWA templates from eleven published MMA studies are attached. Noise reduction is performed by GE Healthcare’s well-established clinical software. MMA templates show absence of noise when the QRS-aligned segments outside the JT window are congruent. Spectral Method analysis does not provide an ECG nor an option to visualize the TWA phenomenon being measured. Finally, noise is not predictive, but MMA-based TWA stratifies risk for cardiac mortality and SCD both in patients with depressed and preserved left ventricular ejection fraction.

3. COMMENT: MMA cannot measure TWA because it does not use specialized equipment, sensors, protocols, noise removal, and interpretation software.
RESPONSE: This statement is not accurate. MMA is FDA cleared to measure TWA with a resolution of 1 microvolt for the purpose of arrhythmic death risk stratification (KO32513). The spectral method’s specialized equipment, sensors, protocols, and interpretation software are not essential to TWA measurement. MMA uses standard electrodes with standard ambulatory ECG or symptom-limited exercise protocols, allowing integration of TWA analysis into the flow of routine clinical evaluation. Interpretation of MMA output in terms of risk is judged by physicians based on the microvolt levels of TWA present in the patients’ ECG; no other interpretation methodology is employed. TWA quantification rather than reporting “negative” or “non-negative” or “indeterminate” test results, as the Spectral Method does, is important because it allows physicians to evaluate the benefit of medical therapy and to track changes in myocardial substrate, such as following acute myocardial infarction.

4. COMMENT: MMA has not standardized which leads to employ, which ECG segments to analyze and reject, cutpoints, and update factors.
RESPONSE: This statement is not accurate. Standards for MMA use were published in in the studies cited below and confirmed in the TWA Consensus Guideline (Verrier et al 2011).

1. Regarding leads for TWA analysis: From 2003, the recommendation has been to employ precordial leads to monitor TWA. Drs. Nearing and Verrier determined in experimental laboratory studies in large animals in 1994 (Nearing et al Cardiovasc Res 1994; 28:1440-49) that precordial leads are optimal for TWA monitoring because they overlie the zone of myocardial ischemia.
2. Regarding ECG segments: Peak TWA levels are generated for each 15 seconds throughout the entire ECG recording after noise reduction is performed by the GE Healthcare software or by users who exclude segments with noise from analysis upon overreading the templates.
3. Regarding cutpoints:
   1. From 2003, MMA has used 47 microvolts to indicate an abnormal level; this cutpoint was set a priori by the standard epidemiological method at 75% of TWA values in patients without events (Verrier et al 2003). The later studies that used this cutpoint are: Nieminen et al (2007, 2014), Stein et al (2008), Ren et al (2012), Hou et al (2013) (incorrectly referred to by Cohen in his public comment as Yu et al 2013), Arisha et al (2013), and Kentta et al (2014).
   2. From 2009, MMA has used 60 microvolts to indicate a severely abnormal level (Minkkinen et al 2009); this cutpoint was set a priori in an exploration of TWA values in a population of >2000 patients. The later studies that used this cutpoint are: Slawnych et al (2009) and Leino et al (2009).
   3. Some studies have used 65 microvolts, which was first explored by Nieminen in 2007 to indicate a somewhat higher level of risk than 60 microvolts. The later studies that used this cutpoint are: Sakaki et al (2009), Maeda et al (2009), Hoshida et al (2012), and Shimada et al (2012).
   4. MMA cutpoints of 47 and 60 microvolts were published in TWA consensus guideline (JACC 2011;58:1309–24) to designate abnormal and severely abnormal heightened levels of risk.
   5. From 2009, MMA has used a 20-microvolt cutpoint to indicate extremely low levels of risk. This cutpoint indicates a hazard ratio =2 for SCD and cardiovascular mortality risk (Minkkinen et al 2009) and a negative predictive value (NPV) of 98% for cardiovascular mortality (Slawnych et al 2009).
   6. The main exception to use of the 47-, 60-, 65- cutpoints for heightened risk, and the 20-microvolt cutpoint for absence of risk was in the REFINE study by Exner et al (2007), who used the non-recommended 5-microvolt cutpoint and the less sensitive update factor of 1/32 and subtracted noise from the TWA voltage. This cutpoint, update factor, and approach to noise have not been used in other studies. The same REFINE data were later reanalyzed with 60- and 20-µV cutpoints and the established 1/8 update factor without subtraction of noise by Slawnych et al (2009), who reported relative risks of 3.4 and 2.5, respectively, for cardiovascular mortality, similar to Exner et al (2007).
   7. A second departure from the cutpoints recommended for MMA was published in a study by Cox et al (2007), who used customized not commercial software with a 64-beat window for averaging and derived an optimized cutpoint of 10.25 µV. This publication is not cited in support of the MMA application.
4. Regarding update factors: The 1/8 update factor has been recommended since 2003 and used by all later clinical studies with the exceptions of a 41-patient study by Cox et al (2007), who used customized software with a 64-beat window for averaging, and the REFINE study by Exner et al (2007), who used 1/32 rather than the established 1/8 update factor.
5. Regarding ranges of values: The Spectral Method output indicated TWA values from 1 microvolt to 10.8±10.0 µV for a maximum value of 20.8 in a study of cardiomyopathy patients (Klingenheben et al 2005). As the extensive averaging applied by the Spectral Method reduces the output TWA value by 3.23 (Hostetler et al 2005), the Spectral Method senses >67 microvolts of TWA, while MMA senses 5-120µV in 2119 patients during routine exercise stress testing (Minkkinen et al 2009).

5. COMMENT: MMA results are not equivalent to Spectral Method results when applied to the same data.
RESPONSE: This statement is not accurate.

1. Exner et al (2007) reported that the Spectral Method generated odds ratios of 2.75 (1.08–7.02, p = 0.034), while MMA resulted in odds of 2.94 (1.10–7.87, p = 0.031) for cardiac death or resuscitated cardiac arrest at 10-14 weeks post-MI. In terms of kappa value, they wrote: “Overall, the agreement between [Spectral Method] and [MMA-based] TWA methods was modest (kappa = 0.17, p = 0.002), but high agreement was observed between the techniques among patients who developed the primary outcome (kappa =0.49, p = 0.009) and those who died (kappa = 0.41, p = 0.02) or suffered a cardiac arrest (kappa = 0.48, p = 0.02).” When the REFINE database was reanalyzed with the standard cutpoints, update factor, and treatment of noise, the relative risk for cardiovascular mortality at the 60-µV cutpoint was 3.4 (Slawnych et al 2009).
2. Hostetler et al (2005) reported a 0.95 correlation between the outputs of TWA voltage by the Spectral and MMA methods in an engineering study with simulated inputs. The output values differed by 3.23 fold, as a result of greater signal averaging by the Spectral Method.
3. MMA is not inferior to the Spectral Method, as Martinez and Olmos (2005) concluded that the nonlinear filtering methods of MMA, which employs recursive averaging, and their own least likelihood ratio “are intrinsically robust to outliers or impulsive noise and have shown to perform better than linear methods in physiological noise.”
4. Selvaraj and Chauhan (2009) criticized MMA but used custom not commercial software with an update factor of 1/16 in comparing Spectral Method to MMA-based TWA.
5. Bini et al (2010) criticized a customized algorithm named “MMAM.” Their findings are not relevant to the commercial MMA software.
6. Abstracts by Costantini, Selvaraj, Sakaki, Oshodi, and Martino and their colleagues, which were cited against equivalence of MMA to the Spectral Method, were never published as full-length manuscripts and thus have not undergone peer review.

6. COMMENT: Spectral Method results cannot be used to support the MMA method. MMA-based TWA should apply for a separate NCD because the current NCD is for the spectral method.
RESPONSE: The current NCD is for T-wave alternans, which is measured by MMA with 1-microvolt resolution equivalent to the Spectral Method, according to FDA document R0 K032513). The current NCD application does not attempt to use Spectral Method data to support the utility and validity of the MMA method. Rather, evidence is provided regarding the equivalence of prediction based on relative risks and other performance parameters including negative predictive value.

7. COMMENT: Dr. Cohen and Mr. DeGreef described the TWA Consensus Guideline (JACC 2011;58:1309–24) as merely a review and implied that the evidence presented for equivalence of the two methods should be disregarded.
RESPONSE: This Consensus Guideline was a collaboration of 11 experts with extensive experience on both the Spectral Method (Drs. Klingenheben, Exner, Hohnloser, Ikeda, Narayan, and Rosenbaum) and MMA method (Drs. Verrier, Exner, Ikeda, and Nieminen) as well as experts in experimental and engineering studies on the TWA phenomenon. The guideline was sponsored by the International Society for Holter and Noninvasive Electrocardiology and cosponsored by the Computers in Cardiology Working Group on e-Cardiology of the European Society of Cardiology, the European Cardiac Arrhythmia Society, and the Japanese Circulation Society. It compared prediction by the Spectral Method in the 7200 patients enrolled in studies with positive results with all MMA-based TWA studies, which enrolled a total of >5,000 patients. Briefly, relative risks for SCD for the Spectral Method were 2.1 to 16.8 and for MMA were 4.4 to 22.6; negative predictive values were similar for the patient groups.

8. COMMENT: In the study by Sakaki et al (2009), investigators were not blinded as to clinical status and outcome.
RESPONSE: We do not know why Dr. Cohen suggests this deficiency as the senior investigator, Dr. Ikeda, is familiar with appropriate study methodology, in fact having published several studies with the Spectral Method (Ikeda et al 2000, 2002, 2006).

9. COMMENT: Hou et al (2013) studied acute post-MI patients and not chronic cardiac patients who are “the main clinical target group for MTWA testing.”
RESPONSE: The acute post-MI period is a time of heightened SCD risk (Solomon et al 2005), when estimation of risk is important.

10. COMMENT: Sulimov et al (2012) found that established cutpoints for peak TWA were not predictive.
RESPONSE: This statement is accurate. However, the authors did not indicate that they had verified TWA levels with reference to the templates or discarded segments with noise. Nevertheless, they reported that MMA-based TWA levels were significantly higher in patients who died suddenly during the followup year and that TWA >53.5 µV was the most significant predictive factor measured for SCD and generated a hazard ratio 5.01 (p=0.005) when heart rate was 100 beats/min. The authors concluded that elevated TWA “significantly increased risk of SCD regardless of LVEF.”

11. COMMENT: Ren et al (2012) did not show MMA-based TWA prediction of cardiac mortality.
RESPONSE: This statement is not accurate. This study reported an area under the receiver operator characteristic curve (AUC) for MMA-based TWA of 0.708 (p=0.021) for predicting cardiac mortality.

12. COMMENT: Shimada et al (2012) did not attempt to predict ventricular arrhythmias or SCD.
RESPONSE: This study reported that MMA-based TWA during the asymptomatic phase of vasospastic angina predicts ventricular arrhythmias with a hazard ratio of 15.5 (p< 0.01).

13. COMMENT: Hoshida et al (2013) did not report MMA’s prediction of arrhythmic death or SCD.
RESPONSE: This statement is not accurate. Hoshida et al (2013) reported a hazard ratio of 5.8 (95% CI 1.6–20.8, P=0.0072) for fatal arrhythmic events in 313 consecutive post-myocardial infarction patients.

14. COMMENT: Green et al (2012) did not find that MMA predicted worse cardiovascular outcomes during followup.
RESPONSE: As indicated in the dossier provided in the public comment by Scott Burke of GE Healthcare, Green et al (2012) employed an inappropriate analysis approach, using mean rather than peak TWA values. It is unclear why the authors analyzed the data incorrectly since GE Healthcare provides a TWA Physicians Guide to users and the study was performed after the publication of the TWA Consensus Guideline (JACC 2011; 58:1309–24), which states the appropriate methodology.

15. COMMENT: Prediction of NSVT shown by Verrier et al (2013) in patients with ST-elevation myocardial infarction (STEMI) does not pertain to risk of lethal arrhythmias or SCD.
RESPONSE: Scirica et al (2010) have recently shown a strong correlation between NSVT and SCD in patients with acute coronary syndrome.

16. COMMENT: Cox et al (2007) showed that MMA was not predictive in the same patients as the Spectral Method.
RESPONSE: Cox et al (2007) employed customized not GE Healthcare’s commercial software with a 64-beat window for averaging ECG data. Nevertheless, MMA-based TWA produced a negative predictive value superior to that of the Spectral Method, due to fewer false-negative determinations in terms of the combined incidence of appropriate ICD therapy, sustained arrhythmias, and all-cause mortality, and achieved a p-value of 0.06.

17. COMMENT: Harvey et al (2008) reported that MMA-based TWA output was strongly correlated with noise.
RESPONSE: These investigators analyzed TWA from limb leads I and II and not from the recommended precordial leads V1-V6 and reported that the majority of output TWA values were 0 to 5 µV, with similar levels of noise. The update factor was not specified. This study was extremely small, enrolling only 6 patients with cardiac disease, and no outcomes were reported. These study limitations raise doubt about the soundness and relevance of this study, which was published in a journal that is unrated in terms of impact factor.

18. COMMENT: The Spectral Method has published rules for analysis since Bloomfield et al in 2003.
RESPONSE: The contention that the Spectral Method analysis rules have been stable is not supported by the facts. Bloomfield et al (2002) described the rules for interpretation by the spectral method as “negative,” “nonnegative,” and “indeterminate.” Indeterminate results occur in 20-40% of tests. This interpretation approach was changed as a result of the study by Kaufman et al (2006) to include “indeterminate due to patient factors,” which includes a component of chronotropic incompetence in patients unable to raise their heart rates to the target range of 105-110 beats/min. “Indeterminate due to patient factors” also includes “transient alternans,” but the duration of this endpoint (seconds, minutes?) was not specified. The “A” and “B” rules in Bloomfield et al (2002) have rarely been referred to in subsequent publications, so that it is unclear whether or not they were actually implemented. The Spectral Method’s initial cutpoint for a negative test was < 1.2 µV (Rosenbaum et al 1994), while < 1.9 microvolts was used in most later studies.

19. COMMENT: MMA studies do not typically analyze SCD risk.
RESPONSE: This statement is not accurate. The following MMA studies established high odds ratios for SCD risk:

• Verrier et al (2003): 4.2 to 7.9 for cardiac arrest or arrhythmic death at 21±8 months
• Minkkinen et al (2009): 4.4 for SCD at 47 months
• Stein et al (2008): 5.5 for SCD at 16.4 months
• Sakaki et al (2009): 22.6 for witnessed SCD at 1.1 years
• Maeda et al (2009): 6.1 for sustained VT or VF at 6 years
• Stein et al (2010): 4.84 for SCD at 4.7 years
• Hoshida et al (2012): 5.8 for SCD at 3.3 years
• Hou et al (2013): 15.07 for SCD at 16±7 months

20. COMMENT: MMA’s NPV is inadequate to rule out ICD implantation.
RESPONSE: This statement is not accurate. MMA’s NPV for SCD is 99%, as reported by Nieminen et al (2007) and Hou et al (2013). This level is equivalent to or better than that of the Spectral Method, which was 97.2% for entire studies as reviewed by Gehi et al (2005) and 95% in the ABCD trial at one year, the most recent large Spectral Method study with a positive outcome (Costantini et al 2009).

21. COMMENT: MMA does not predict appropriate ICD discharge but the Spectral Method does, citing Molon et al (2013).
RESPONSE: As Dr. Cohen is fully aware, TWA studies conducted with the Spectral Method were not found to predict appropriate ICD discharge in the MASTER (Chow et al 2008) or SCD-HeFT (Gold et al 2008) trials. He and his colleagues attempted to explain the absence of prediction by the Spectral Method in studies in which ICDs were in high use (Hohnloser et al 2009), questioning the accuracy of appropriate ICD discharge as an endpoint because arrhythmias that may be self-terminating can trigger discharges. Also, the devices themselves may trigger firing (Germano et al 2006). Therefore, prediction of SCD in primary prevention patients has been deemed a more accurate test of the TWA algorithms than appropriate ICD discharge. MMA’s prediction of appropriate ICD discharge has not been analyzed, but MMA achieves success in SCD risk stratification with hazard ratios and negative predictive values that are equivalent or superior to the Spectral Method. Finally, the letter to the editor by Molon et al (2013), cited in support of Spectral Method, is not a peer-reviewed publication.

In summary, based on a sizeable body of evidence in studies enrolling >5000 patients, we believe that use of MMA-based TWA for estimation of risk for SCD, which takes a toll of 310,000 Americans annually (Go et al 2014), predominantly in the Medicare age group, as the median age is 69 (Reddy et al 2009), is reasonable and necessary in Medicare patients as it accurately measures TWA and offers a valuable test to the CMS population, with considerable evidence to support its utility in identifying individuals at risk for SCD, thereby aiding physicians in evaluating appropriate care.

Sincerely,

REFERENCES

Images [PDF]

Less

Less