Should Guidelines Support TAVR for Low-Risk Patients? Two Views

It’s been a full year since the US Food and Drug Administration expanded use of TAVR to patients with severe aortic valve stenosis at low surgical risk, but reaching a full endorsement via guideline recommendations is not a simple matter, as two opposing arguments make clear in the Journal of the American College of Cardiology today.

“We need to have an explicit conversation with our patients that we simply at this point do not have any evidence supporting durability of outcomes, which is what is key for patients at lower risk,” Sanjay Kaul, MD (Cedars-Sinai Medical Center, Los Angeles, CA), the author of one of the papers, told TCTMD.

Conceding the point that durability still needs to be established, Michael J. Mack, MD (Baylor Scott & White Health, Dallas, TX), and David H. Adams, MD (Mount Sinai Health System, New York, NY), insist nevertheless that it is “unjustified and impractical to withhold novel valve therapies for a decade.” One solution, they propose in a counterpoint article to Kaul’s, would be for guideline committees and other “stakeholders” to stress the need for longer-term follow-up by withholding a Class I recommendation for use until more time has passed.

In August 2019, based on the results from  Evolut Low-Risk and PARTNER 3 (for which Mack was co-principal investigator), the FDA expanded the indication for TAVR to patients with a 30-day surgical mortality risk of < 3% to 4%. TAVR was superior to SAVR in PARTNER 3 at 1-year follow-up for the outcome of death, any stroke, or rehospitalization, and it was noninferior to SAVR in Evolut Low-Risk at 2-year follow-up for the outcome of death or disabling stroke. Not yet updated, the 2017 American College of Cardiology/American Heart Association focused update on valvular heart disease continues to recommend SAVR as the preferred therapy for low-risk patients.

A Critical Appraisal

In his review of the evidence published today, Kaul concludes that the current level of evidence is insufficient to make a Class I guideline recommendation for TAVR in low-risk patients. He points out that compared with intermediate-risk patients, the number of low-risk individuals in the pivotal trials was smaller, as was the total number of primary endpoint events, which were nearly threefold lower. Similarly, the total number of deaths or disabling strokes at 1 year in the low-risk cohorts was between four- and sevenfold lower in comparison with intermediate-, high-, or inoperable-risk patients.

Kaul also cites a number of other issues, including less-robust trial design for the low-risk category than higher-risk categories, differences in key outcomes among the pivotal low-risk trials despite similar baseline CV risk, a steep gradient in the 1-year risk of death or disabling stroke between the TAVR arms in PARTNER 3 and PARTNER 2 trials, and concerns about long-term outcomes based on 5-year follow-up results of PARTNER 2.

“We have the PARTNER 2 data at 5 years with more than doubling of the primary endpoint events [compared to 2-year follow-up], and there seems to be an uptick in death or disabling stroke event rate with TAVR,” Kaul observed to TCTMD. “One could argue that as the number of events doubles, the noninferiority margin that would be considered acceptable should narrow. So, it does not meet the criteria for noninferiority at 5-year follow-up.

“Similarly, we see an uptick in events, particularly thrombotic events, in the PARTNER 3 trial at 2 years,” he continued. Indeed, in a presentation earlier this year at the virtual American College of Cardiology 2020 Scientific Session, investigators reported significantly more valve thromboses in the TAVR cohort compared with the SAVR cohort between years 1 and 2, with a total incidence at 2 years of 2.6% versus 0.7% (P = 0.02).

Kaul said the thrombosis concerns are particularly worrisome because not only is there no way to predict which low-risk patients might develop it, the benefit-risk balance becomes even more adverse because there are no effective treatments to mitigate that risk.

Another important issue, according to Kaul, is that in Evolut Low-Risk, fewer than 10% of patients completed the 2-year follow-up, requiring model-based imputation for the remaining 91.2%.

According to Kaul, a minimum of 5-year follow-up might be necessary to achieve a sufficient number of hard endpoints such as deaths or disabling strokes to justify a guideline recommendation for TAVR in low-risk individuals.

Some Agreement, Some Rebuttal

As for Kaul’s other concerns, Mack and Adams write: “It is indeed reasonable to ask what can be done to address the points raised by Kaul regarding determining a sufficient evidentiary bar to change guidelines. What are some alternatives to the current clinical trial ecosystem that can generate evidence in a timely, cost-efficient manner to facilitate prompt introduction of novel therapies into clinical practice while attaining the scientific rigor that Dr. Kaul is requesting?”

What’s needed, they say, is a balance struck between timely evidence generation at a reasonable cost, while also “generating meaningful findings that are likely to impact clinical practice.”

Worth noting, they add, is that the sponsors of both low-risk trials have agreed with regulators to follow patients out to 10 years. “Perhaps Kaul is right to temper the enthusiasm” until more is known, Mack and Adams say. Going forward, particularly given the criticisms that a number of transcatheter valve and coronary revascularization trials have received, it would make sense for more collaboration between guideline committees, regulators, and payers at the time trials are being designed, they propose. “If we have learned anything from the controversies expressed around device trials in our field, it is that the acceptance of answers one obtains from a trial depends to a large degree on how the question is asked at the outset, and the statistical plan that will help answer it.”

As for Mack and Adams’ specific suggestion that guidelines skip the Class I recommendation until longer-duration follow-up is in, Kaul said that he is inclined to agree that a Class IIa or b recommendation would be more appropriate, “provided that patients are told that we still don't have long-term data regarding durability or the consequences of left bundle branch block or paravalvular regurgitation or conduction abnormalities requiring pacemaker.”

Asked to weigh in on the debate for TCTMD, B. Hadley Wilson, MD (Atrium Health Sanger Heart & Vascular Institute, Charlotte, NC), said Kaul, Mack, and Adams all make valid points that highlight the importance of shared decision-making and full disclosure.

“In taking care of patients, we need to be clear that the ultimate results and outcomes are not known in finality, and so it is possible that later on these recommendations and guidelines might be changed,” he said. “Patients should be aware that they may be undergoing a therapy that later on actually may not be the one that is favored by the guidelines or even FDA. Is that too much put on our patients? I don't think so, because patients want to be part of the decision-making. In many cases, patients are willing to take low risks, or perhaps even consider less-[favorable] outcomes in order to achieve a therapy which suits their particular lifestyle or their particular health and age at that time.”