Well, well. Predicting clinical trial results is as humbling as taking care of patients.
At the Transcatheter Cardiovascular Therapeutics (TCT) 2018, Gregg Stone, MD, from Columbia University, New York City, presented the astonishing results of COAPT, the second randomized controlled trial comparing MitraClip (Abbott) with medical therapy alone for secondary mitral regurgitation (MR) in patients with heart failure.
In COAPT, the MitraClip performed far better than known winners in cardiology, such as percutaneous coronary intervention (PCI) for acute myocardial infarction or implantable cardioverter-defibrillators (ICDs) for heart failure. The number needed to treat (NNT) to prevent one hospitalization for heart failure at 24 months was 3, and the NNT to prevent one death at 24 months was 6. In other words, the device looked almost as good as clean drinking water, sanitation, and antibiotics.
That's strange because 1 month ago, French investigators reported in the Mitra-FR trial that the same percutaneous device that approximates edge-to-edge repair had no incremental benefit for heart failure hospitalization or death in similar patients
First, the findings of the COAPT, then some comments.
In short, COAPT investigators randomly assigned slightly more than 600 patients with heart failure and significant secondary MR who remained symptomatic despite maximal medical therapy to either the device plus medical therapy or medical therapy alone. The primary endpoint was heart failure hospitalizations at 24 months.
Six Key Results
The rate of heart failure hospitalizations was 36% per patient-year in the device group vs 68% per patient-year in the control group (hazard ratio, 0.53; 95% confidence interval [CI], 0.40 - 0.70; P < .001).
Complications from the nearly 3-hour procedure occurred in only 3.4% of patients.
The device worked as planned. MR grade 2+ or lower at 12 months occurred in 199 of 210 (95%) patients with the device vs 82 of 175 (47%) in the control group.
Patients in the device group scored higher on quality-of-life surveys, did better on 6-minute walk tests, and more often achieved New York Heart Association functional class I or II.
Death from any cause at 12 months occurred in 57 patients (19%) in the device group vs 70 patients (23%) in the control group (hazard ratio, 0.81; 95% CI, 0.57 - 1.15, which met criteria for noninferiority.
Death from any cause at 24 months strongly favored the device group, at 29% vs 46% in the control group (hazard ratio, 0.62; 95% CI, 0.46 - 0.82).
COAPT was a clear winner. I spoke with Sanjay Kaul, MD, from Cedars-Sinai in Los Angeles, California, at the meeting, and he said there is no other way to describe COAPT except as a positive trial. He noted the obvious: The improved outcomes with the device aligned with its ability to reduce MR. That lends plausibility to the findings.
Another positive for COAPT was the balanced baseline characteristics in the two arms. I sent the results to John Carlisle, MD, the UK doctor whose methodology exposed the PREDIMED trial for its irregular randomization. He confirmed that the baseline continuous variables for both COAPT and MITRA-FR trials were normal and consistent with the allocation methods.
What we have then is two randomized trials with conflicting results. That means we don't know whether the device works, and we need a tie-breaker trial. Fortunately, one is coming soon. Kaul told me to expect results of RESHAPE-HF2, a European trial of about 400 patients, in the next few months.
Spin Control and Nuance
COAPT and MITRA-FR had slightly different enrollment criteria, but patient selection does not explain the disparate results. While there were more severe MR, less variable medical therapy, and fewer complications in COAPT, that's not sufficient to account for one trial being negative yet the other turning in mortality results threefold better than seen in ICD trials.
The proof of similarity of patients in the two trials is clear: One-year mortality rates in the control groups were 23% in COAPT and 22% in MITRA-FR.
The astonishing 17% absolute reduction in death in COAPT deserves comment. One point is that MITRA-FR had 1-year follow-up and COAPT went for 2 years. So it's worth looking at overall death rates at 1 year. My colleague and frequent coauthor, Andrew Foy, MD, from Penn State, helped me sort this out.
In COAPT, all-cause death at 1 year was 19.1% in the device group vs 23.2% in the control group. That difference generated a hazard ratio of 0.81 (95% CI, 0.57 - 1.15); note that the CIs include 1.0. MITRA-FR reported all-cause death at 1 year as 24.3% vs 22.4% in the device and control groups, respectively; their hazard ratio was 1.11 (95% CI, 0.69 - 1.77), again overlapping 1.0. Both trials, therefore, found little difference at 1 year. Foy explained that when you do a Forrest plot of these numbers, the confidence bars overlap and the heterogeneity is low, indicating the findings align with each other.
Things change a lot when you compare all-cause death of the 2-year data from COAPT against the 1-year data from MITRA-FR. In COAPT, all-cause death in the device group was 17% lower in absolute terms, and the hazard ratio plummeted to 0.62 (95% CI, 0.46 - 0.82), which is highly significant. Foy explained that on a Forrest plot, the heterogeneity goes way up. Translation: the difference in death rates was driven by the 2-year results of COAPT.
One conclusion is that survival differences from MR reduction with the device accrue over time. The problem with making that leap is missing data in COAPT. Stone mentioned this in the presentation, and the authors wrote in the manuscript that withdrawal from the trial was more frequent in the control group. Missing data increased as the trial went on, decreasing the numbers of "eligible" patients over time.
Note that this critique is similar to Dr Milton Packer's criticism of the CASTLE-AF trial of atrial fibrillation ablation in patients with heart failure—namely, that reduction in the number of eligible patients at the tail end of trials reduces confidence in the results. That makes the similar all-cause death results at 1-year more reliable.
An astonishingly positive trial in the face of a negative trial forces one to consider bias. In COAPT, and most device trials, patients and clinicians knew the treatment assignment. Performance bias occurs when investigators treat device patients more effectively; ascertainment bias occurs when knowledge of the treatment arm affects clinical decisions, such as the decision to send a patient to the hospital for symptoms of heart failure.
Patients were enrolled in COAPT by a 1:1 randomization protocol stratified by cardiomyopathy type and site. US Food and Drug Administration reviewers and authors of systematic reviews on this topic will also probe the vigor of allocation concealment, which is an important control for selection bias (eg, healthier patients in the device group or sicker patients in the control group). Remember, allocation concealment is not the same as randomization.
Percutaneous mitral valve repair is expensive and invasive. The conservative practitioner considers the totality of the evidence for percutaneous mitral valve repair, sees no answer, and thus waits for more data.
The scientist, however, gets a reprieve. If COAPT had been negative, the percutaneous approach to secondary MR may have been dead. If RESHAPE-HF2 confirms COAPT, we will have to rethink the long-held view on the physiology of secondary MR. Maybe the problem isn't just the failing ventricle.
What makes clinical science and medical practice so beautiful, so enthralling, is the learning. And what better way to learn than making mistakes, and having to let go of strongly held views. Then again, if RESHAPE looks more like MITRA-FR, we will have to live with uncertainty, which is another upside of the practice of medicine.
© 2018 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: John M. Mandrola. MitraClip for Secondary MR -- The Redux - Medscape - Sep 23, 2018.