Q&A: Cardiologist and EARLY TAVR co-author questions study’s execution

What was your experience enrolling and treating patients for EARLY TAVR? Any big takeaways?

We were a top-10 enrolling site and had some interesting patient encounters. Some of these folks wanted TAVR right away and actually came to us fully aware of the procedure. And some of them were disappointed when we told them they weren’t commercially indicated for TAVR, even when I explained there was no proven benefit to treatment in asymptomatic patients. 

Patients were thus very interested when we told them about EARLY TAVR, because many saw it as their only chance of undergoing TAVR. I explained to all of them that they would either be randomized to the TAVR arm and undergo treatment, or they would be randomized to the clinical surveillance arm and have their symptoms closely monitored. Each patient then talked it over with me, our surgeons, our valve clinical coordinator, our research nurses and others—and if they wanted to participate, they signed the dotted line.

One thing I noticed was that some patients randomized to the clinical surveillance arm immediately started developing mild symptoms. I’m not saying that this was intentional, but they did know they would likely be treated with TAVR—the procedure they wanted from the jump—as soon as those symptoms started to develop. And once the patient reported symptoms in EARLY TAVR, that was basically it; they were going to get a TAVR valve. It was not up to a screening committee or an objective test, and there was no waiting to see what happened next. As soon as they were symptomatic, we performed shared decision-making and went forward with the procedure.

Overall, though, we saw a lot of variability with regards to when patients in the clinical surveillance arm developed symptoms and underwent TAVR. Our average time to treatment for that arm was 444 days at our site, which appears to be much longer than other sites in the study; some patients went years before treatment. 

Did you help write any of the text for EARLY TAVR or play any other additional roles?

No. I was part of the review process, but I, and most others, did not get much time. I literally had about 48 hours to read and analyze the manuscript. During that perusal, I could instantly tell I did not agree with the direction my co-authors had gone with these data—but I could also tell they weren’t going to change those things based on my input. It would have been a waste of time, to be honest with you. They already knew they were aiming for The New England Journal of Medicine and simultaneously presenting the data at TCT—no one was going to pump the brakes on that. And, again, there was such a short review period that I didn’t get to fully digest how I felt at the time. 

I was proud to be a significant part of EARLY TAVR. We enrolled really well and did not game the trial in any way. But that doesn’t mean I cannot have a distaste for how it was all portrayed in the final manuscript. 

Let’s get into that distaste now. What were your biggest issues with EARLY TAVR? 

There are a few things to discuss here. First, as you know, patients in EARLY TAVR landed in either the treatment arm and underwent TAVR pretty much right away, or they landed in the clinical surveillance arm and were monitored closely. They called it a conversion when those clinical surveillance patients went on to undergo TAVR due to symptoms, but that should have been viewed as a crossover. When you have one arm being treated with something that the other arm can then be treated with, they are not a convert; that’s a crossover. 

So one of the big issues I had with the manuscript was the wording, with the emphasis on these conversions. They were not converted. TAVR was the near monolithic exit strategy for these patients. That impacts the statistical analyses of this trial significantly.

Another thing I want to highlight is that, looking at the totality of the data. there was no mortality benefit for TAVR in asymptomatic severe AS. The benefit was seen with the study’s composite endpoint, but that was driven by unplanned hospitalizations, specifically a sponsor-devised endpoint that included elective AVR—in this population, mainly TAVR—out to six months post-randomization in the clinical surveillance arm. 

There are a lot of objectionable things here. And I want to talk about subtraction anxiety next, because that seems to be the biggest reason a lot of these clinical surveillance patients ended up getting TAVR. 

Subtraction anxiety—can you define that for me?

It’s the same basic concept as FOMO, that acronym that stands for “the fear of missing out.” If I had to explain this to my 11-year-old, I’d tell her these patients had FOMO.

Patients, in large part, did not participate in EARLY TAVR because they wanted to contribute to medical science, right? Many participated because they wanted a TAVR. When patients landed in the clinical surveillance arm, though, how do you think they felt? They are naturally going to be wishing they had landed in the other arm. They are going to have subtraction anxiety.

You can see it in the Kansas City Cardiomyopathy Questionnaire (KCCQ) scores, which help track a patient’s quality of life. Patients who “converted” to TAVR in the first three months had the absolute worst KCCQ scores; their lives just went to complete trash. Why is that? Well, it’s because of subtraction anxiety. This is actually exactly what subtraction anxiety looks like.

Again, to be clear, I don’t want to say these patients were faking anything. But this is a very real psychological phenomenon and it truly made them feel worse after randomization. I think the worst harm done by EARLY TAVR was randomizing patients to the clinical surveillance arm. That randomization may have deteriorated KCCQ scores more than anything else.

What should have been done instead?

There are a couple things. First, surgical aortic valve replacement (SAVR) was not a part of this study—but if it had been, that may have really helped distill that subtraction anxiety. Patients do not see the invasiveness of SAVR as holding the same benefits as percutaneous transfemoral TAVR. If you tell them they may be treated with SAVR or TAVR, perhaps randomized, when they develop symptoms, I think that could have changed things.

Also, as a side note—maybe there is a really great benefit for early SAVR for asymptomatic patients. We could have found out if we made SAVR a part of this study.

The other thing that could have maybe been done differently is, you could have included a sham control. You bring some patients into the lab/OR, you make them think they are undergoing TAVR—but they don’t get the valve. This is a pretty extreme option, but it could have helped us learn more about all of this.

With all of that said—what should interventional cardiologists take away from EARLY TAVR?

Well, we should be surveilling our patients. Some patients who we thought were asymptomatic were excluded from EARLY TAVR after their screening treadmill test, because we found that they did, in fact, have symptoms or a symptom equivalent. That is a big deal. It shows the importance of performing these treadmill tests. It’s a good reminder of how important it is to thoroughly evaluate these patients.

If we could have somehow completely avoided the subtraction anxiety, the FOMO, I think performing treadmill tests annually on the patients in the clinical surveillance arm would have been a good idea. 

What lessons do you take away from this experience?

Unfortunately, in structural heart disease, we have fallen down a slippery slope over the last few years when it comes to clinical trials being sponsored, executed, and marketed by industry. The good thing is, you see a lot more people talking about this influence now than ever before, and I think that critical thought and push for data transparency should help lead to big improvements in the future. There will be more noise. 

At the end of the day, industry deals with a different value system than physicians or patients. Bias presents in this context no matter what we do, but we need to minimize it to the extent we can.

I understand that industry plays a large role in our ability to amass and roll out scientific data, but we have to give all sorts of people voices at the table, even when those voices are critical. There is part of me that wishes I had spoken up to the co-authors at the time when I first saw the EARLY TAVR manuscript, but when the cards are played before the hand is dealt, there is no point. Hopefully, we can establish more inclusive, dynamic and collaborative approaches in the future.