AUTHEARTVISIT Reignites RCT vs Real-world Data Debate for AV Repair

VIENNA, Austria—A large observational study of patients with severe aortic stenosis has raised eyebrows—and no shortage of skepticism—with data suggesting that surgery is associated with superior survival when compared with the transcatheter approach.  

In a data set capturing most of the aortic stenosis procedures performed in Austria, the probability of survival over a median follow-up of 4 years was roughly 50% higher among patients treated with surgical aortic valve replacement, both in the overall population and in a propensity-matched analysis, Johann Auer, MD (Krankenhaus St. Josef, Braunau am Inn, Austria), reported last week during a late-breaking science session at the European Association for Cardio-Thoracic Surgery (EACTS) annual meeting.

“In the first year, there was no difference in mortality between patients selected for SAVR versus those selected for TAVR,” said Auer. “However, beyond 2 years, there was a survival advantage for patients undergoing surgical bioprosthetic aortic valve replacement.”

While the randomized controlled trials have firmly established a role for TAVI in patients with symptomatic severe aortic stenosis across the surgical risk spectrum, Auer said those studies should be appraised cautiously. “Randomized trials are usually the gold standard for establishing the efficacy of medical interventions, but they can’t capture every aspect of the effect of a therapeutic intervention in clinical practice,” he said.

Several physicians pushed back against those conclusions. Cardiac surgeon Volkmar Falk, MD (German Heart Center, Berlin), said that while exploring such real-world data can be useful, “it’s often not risk-adjusted and it’s very difficult to [move] from insurance data to quality outcome data.” Physicians, he stressed, should be very cautious about making clinical decisions based on registry data, given their inherent limitations.

Senior investigator Hendrik Ankersmit, MD (Medical University of Vienna, Austria), who spoke during an EACTS session devoted to bias in randomized trials, stressed that their analysis is unique in that they captured real-world mortality outcomes using data from the national health insurance registry.

“When we look at the patient data, the actual data we get from the Austrian insurance [registry], TAVI is completely a hazard to all patients,” he said. “I would encourage everybody to look within their own country, within their own insurance [registries], and get these data. I think it’s an additional way to interpret the results we get from all these industry-driven trials.” 

However, Cristian Baeza, MD (University Hospitals, Cleveland, OH), a cardiothoracic surgeon who also performs TAVI procedures, struggled to make sense of the data, telling TCTMD that his sense is that the TAVI patients are simply sicker and more fragile than those sent for surgery, and this risk isn’t picked up in the statistical adjustments.

“Why did these patients die?” he asked, adding that he’s asking this even as a surgeon who wants to keep doing SAVR. “I don’t think they died just because they had TAVI,” he said. “It’s not directly related to the procedure.”

Benoy Shah, MBBS, MD (University Hospital Southampton NHS Foundation Trust, England), past president of the British Heart Valve Society, also was skeptical. “The problem with this is that physicians and surgeons were left to decide who should have TAVI and who should have surgery,” he told TCTMD. “We don’t get the granularity about what went into that decision. Certain patients might have certain comorbidities or frailty which pushed them one way or the other.”

Moreover, said Shah, the observational data contrast with midterm mortality data from multiple randomized trials in intermediate- and high-risk patients, such as PARTNER 1, CoreValve US High-Risk Pivotal trial, PARTNER 2A, and SURTAVI, in which there were no significant differences in the risk of death between the two procedures at 5 years.

A few weeks from now, at TCT 2023, investigators will present 5-year clinical and echocardiographic data from PARTNER 3, which compared TAVI versus SAVR in low-risk patients, and 4-year outcomes from the Evolut Low-Risk Trial. 

Valve Replacement Over 10 Years

The Austrian observational study, known as AUTHEARTVISIT, drew on data from the national health insurance fund, representing 95% of the population. In all, 18,882 men and women with severe aortic stenosis age 65 years and older who were treated between 2010 and 2020 were included. Of these, 62.2% were treated with surgery and 37.8% underwent TAVI. Investigators excluded patients with coronary revascularization within 4 months of valve replacement, those undergoing concomitant heart surgery, and surgery patients treated with mechanical valves.

In the overall population, surgery was associated with a 55% higher probability of survival (HR 1.55; 95% CI 1.47-1.64), a benefit that was maintained in the propensity-adjusted analysis (HR 1.51; 95% 1.40-1.63). Restricting the analysis to those age 75 years and older, survival was again better with surgery in both the overall and propensity-matched analysis. There was no difference in the risk of reoperation, new-onset heart failure, MI, or stroke between the two interventions, but the risk of permanent pacemaker implantation was higher with TAVI (11.2% vs 4.5%; P = 0.02).

Sabine Bleiziffer, MD (Heart and Diabetes Center NRW, Bad Oeynhausen, Germany), who moderated the late-breaking session, said the Austrian data mirror findings from the German Aortic Valve Registry (GARY), which her group published in 2021. Among 18,010 patients with symptomatic severe aortic stenosis treated in 2011 and 2012, all-cause mortality in a propensity-matched analysis was significantly higher with TAVI than SAVR after 5 years of follow-up.

However, “our conclusion was that our heart team were doing a really good job selecting patients for each of the therapies,” said Bleiziffer, noting the TAVI-treated patients likely had higher risk uncaptured with propensity matching that pushed them towards that procedure.

Auer acknowledged the limitations of their observational research, noting that choice of intervention is not the only factor that might be influencing survival. In the multivariable analysis, age, sex, prior heart failure, and chronic kidney disease at baseline were also associated with survival in the overall and propensity-matched models. In fact, frailty couldn’t be captured from the Austrian insurance data and couldn’t be accounted for in their analysis.

Despite those drawbacks, though, “we have to acknowledge that these are the data,” said Auer.

While propensity-score matching is one way to overcome the limitations of observational research, it’s only an attempt to make observational data more like randomized trials, said Shah. “I’m personally not a big fan of propensity-score matching,” he said. “Why not just do the randomized trial? If this was a cohort where you were never going to get a randomized trial, and this was the best you could get, then fine, go ahead. But that’s not the case.”

Cardiac surgeon Victor Dayan, MD, PhD (Centro Cardiovascular Universitario, Montevideo, Uruguay), wasn’t as concerned about the risk of confounding in the retrospective study. He noted that while the randomized trials haven’t shown a mortality difference in midterm follow-up, the survival curves did cross to favor surgery at 2 years in PARTNER 2A.  

Randomized Trials Take Heat

At EACTS, surgeons also reiterated concerns they’ve raised before about the SAVR/TAVI randomized data. Fabio Barili, MD, PhD (S. Croce Hospital, Cuneo, Italy), challenged the external validity of the randomized trials, noting that many have methodological issues that may increase the risk of bias, a concern he and others have highlighted previously. For example, there are differences in the loss to follow-up, particularly among surgical patients, and use of additional procedures, such as higher rates of concomitant coronary revascularization with surgery. There are also deviations from the randomly assigned treatment, noted Barili.

Austria is roughly in the middle of the pack in Western Europe when it comes to the number of TAVI procedures performed per 1 million inhabitants, according to data presented elsewhere at EACTS 2023. Thierry Folliguet, MD, PhD (Assistance Publique Hôpital de Paris, France), who presented the data, reported that there were 152 TAVIs per million patients performed in the last quarter of 2022, the most recently available data. This is less than in Germany, France, and Switzerland, among other countries, but higher than in Sweden, the Netherlands, and the United Kingdom.