New Absorb BVS Meta-analysis Begs the Question: When to Stop Studying a First-Gen Device?

A new, large meta-analysis should vanquish any lingering hopes that the current iteration of the Absorb bioresorbable vascular scaffold (BVS) may yet prove safer, more effective than, or even equivalent to a modern-day metallic DES, at least in the short-term, experts say.

Some also say that this means the first-generation Absorb scaffold shouldn't still be under study, period.

In this latest meta-analysis, which included more than 5,500 patients randomized to either the Absorb BVS or the Xience everolimus-eluting stent (EES; both Abbott Vascular), patients who received Absorb saw a 52% increase in the risk of myocardial infarction and a 40% increase in ischemia-driven repeat procedures in the target lesion. The risk of scaffold thrombosis with Absorb was threefold higher over the 2-year period and a sobering tenfold higher between years 1 and 2.

“What we see is that honestly, beyond any reasonable doubt, this hasn’t worked out yet,” Peter Jüni, MD (University of Toronto, Canada), who was not involved in the study, observed to TCTMD. “The device that we have here, for various reasons, just isn’t ready for clinical use because of both safety and efficacy issues, as compared with an admittedly excellently performing metallic DES.”

That’s a conclusion the authors themselves—including principal investigators for the ABSORB II and III trials, ABSORB China, and ABSORB Japan—appear to agree with.

What we see is that honestly, beyond any reasonable doubt, this hasn’t worked out yet. Peter Jüni, MD

“Compared with metallic EES, BVS have higher 2-year rates of the device-oriented and patient-oriented composite endpoints, which are driven by an increase in rates of target vessel-related myocardial infarction and device thrombosis with BVS during the 1-year to 2-year follow-up,” write Ziad Ali, MD, DPhil (NewYork-Presbyterian/Columbia University Medical Center, New York, NY), and colleagues in an early online publication in the Lancet this week.

Those conclusions may come as no surprise to people who have been following the troubles faced by the first-of-its-kind “dissolving” stent in recent years. But as Jüni writes in an accompanying editorial co-authored with Raffaele Piccolo, MD (Bern University Hospital, Switzerland), and Stephan Windecker, MD (Bern University Hospital), previous underpowered analyses unearthed adverse event rates with the Absorb BVS that, at best, were “borderline significant.” With the added numbers in the current analysis, the editorialists write, the weight of the evidence goes from “weak to moderate” to “strong to overwhelming.”

Absorb gained regulatory approval in countries around the globe before being restricted to research-only settings in Europe and Australia. It was also the subject of a strongly worded Food and Drug Administration safety alert issued to doctors in the United States.

ABSORB Investigators Unite

Ziad et al included seven studies in their meta-analysis: ABSORB II, ABSORB III, ABSORB China, ABSORB Japan, EVERBIO II, TROFI II, and AIDA. Both ABSORB III and AIDA were powered for composite clinical endpoints, they note.

The device-oriented composite endpoint (cardiac mortality, target vessel-related MI, and ischemia-driven TLR), as well as the individual endpoints of target-vessel MI and TLR alone, were all significantly higher in the BVS group. Cardiac mortality, however, was not statistically different between groups. Definite or probable stent thrombosis—the primary safety endpoint—was also significantly greater in the Absorb group.

Meta-analysis: Key Findings

A landmark analysis at the 12-month mark showed that events continued to be significantly higher for the BVS-treated patients, both for the composite device-orientated endpoint and for device thrombosis, which occurred in 11 BVS-treated patients between years 1 and 2 but in none of the EES-treated patients.

In interviews with TCTMD, both Ali and senior author Gregg Stone, MD (New-York Presbyterian/Columbia University Medical Center), pointed to a number of unique aspects of the current meta-analysis including its inclusion of the as-yet unpublished 2-year results from ABSORB III, a patient-level pooled analysis (almost 3,400 patients), and the range of subanalyses performed.

“What differentiates our study is that our meta-analysis is not only comprehensive, but also unique in that it includes individual patient data from the four Absorb trials, allowing us to look at the performance outcomes in specific subgroups and look at predictors of device failure,” Ali said in an email to TCTMD. “We identified that preprocedure reference vessel diameter less than 2.25 mm predicted 2-year device and patient-oriented composite endpoints, driven by MI, target-vessel MI, ischemia-driven target vessel revascularization, and device thrombosis.”

Also unique, said Ali, is an analysis showing that when device thrombosis events are excluded, the BVS and EES “otherwise perform similarly, suggesting that specific attention needs to be paid to minimize the risk of device thrombosis, either procedurally or through patient selection.”

Everything is pointing in the same direction. Joanna Wykrzykowska, MD

Commenting for TCTMD, Joanna Wykrzykowska, MD (Academic Medical Center, Amsterdam, the Netherlands), lead investigator for the AIDA trial, said that the new meta-analysis confirms a range of other meta-analyses that have previously been published in many of the top-tier cardiovascular medicine journals. “Everything is pointing in the same direction,” she said. Her chief concern, however, is the fact that the analysis only goes out to 2 years. “It’s the very late events which are so very unpredictable,” she said.

And because the patient-level data were taken only from the ABSORB series of trials, that means “simple lesions and simple patients,” Wykrzykowska observed. As such, she’s not convinced that the predictors identified by Ziad et al related to vessel sizing and operator technique would hold up in more complex patients.

Small Absolute Differences?

Stone, for his part, pointed to the fact that the absolute difference in event rates between the BVS- and EES-treated patients is actually very small.

“Clearly this meta-analysis has recapitulated the results of others showing that Absorb is associated with a higher adverse event rate, target lesion revascularization, target-vessel MI, and scaffold thrombosis in particular,” Stone told TCTMD. But importantly, he continued, “the difference in event rates is relatively modest, so approximately a 2% difference in TLF and a 1% absolute difference in scaffold and stent thrombosis. So a number needed to treat of approximately 50 and 100. And while that shouldn’t be underestimated—the importance of those to individual patients—what it means, for example, for target lesion failure is that 49 out of 50 patients have the same 2-year results with Absorb and Xience and 1 out of 50 patients don’t.”

That’s key, he added, because these trials were done with a relatively thick-strut, early-generation device and suboptimal technique.

Stone continued: “I think it’s not at all inconceivable to think that a better device and improved technique is going to eliminate those early differences, which will allow what will hopefully be shown to be long-term advantages of no longer having the permanent metal frame in the vessel, which should appear after the time of complete bioabsorption, which is 3 years.”

The editorialists, however, use the term “number needed to harm,” concluding that the ones seen here are far too low. “It cannot be anticipated that the potential long-term benefits of the current version of Absorb BVS will offset the excess in adverse events reported during the first 2 years,” they write.

But Stone says he “strongly disagrees” with the conclusion that longer-term outcomes, even with the current-generation device, won’t offset these early events.

“The absolute differences in risk, depending on what event you’re talking about, are one or two out of a hundred patients,” he told TCTMD. “And we know that metallic DES from years 3 to 10, and probably for the life of the patient, have about a 2-3% per year rate of TLF. So if Absorb or a bioresorbable scaffold can capture back a 1% or a half percent of those events on a per year benefit, then it won’t be many years at all before it makes up that early difference.”

That being said, he noted: “We certainly would like to have a scaffold that is as safe in the early stage as the metallic drug-eluting stent, but I do think that [the current] scaffold can make up even the difference you’ve seen in that meta-analysis.”

Looking Ahead, Not Back

Abbott does have a next-generation device in the works, Stone noted, code-named “Falcon” for now. The device, still investigational, has a strut thickness of less than 100 µm and improved expansion characteristics, he said. By way of comparison, the current-generation Absorb BVS is 150 µm. TCTMD has also reported on a range of other devices being developed by other manufacturers that have demonstrated promising early results, albeit in small numbers of patients.

Moreover, insights into the importance of optimized technique, sizing, and patient selection will come from ABSORB IV—30-day results are expected at the TCT 2017 meeting. In their editorial, however, Piccolo et al write that the true value of specific implantation techniques, avoidance of small vessels, and prolonged dual antiplatelet therapy “is subject to debate.”

Indeed, they say, it has been estimated that only half of patients with late or very late scaffold thrombosis might benefit from an optimized implantation technique, “which is unlikely to address adverse events related to late discontinuity, recoil, and neoatherosclerosis.”

This Stone characterized as “educated speculation,” saying, “I don’t think there is any way to come up with that estimate.”

In their meta-analysis, Stone noted, most of the difference in scaffold/stent thrombosis occurred in vessels less than 2.25 mm by quantitative coronary angiography. Moreover, he added, the difference in scaffold/stent thrombosis was also “strongly mitigated by routine high-pressure postdilatation.”

That’s all post hoc data, he acknowledged, “and I do agree that optimal technique is not going to eliminate all the risks.” That said, Stone continued, “I do think what most people would agree with is that this is going to take a combination of optimal scaffold-specific technique plus an improved scaffold from this first-generation device if we’re going to have a bioresorbable scaffold that that is as safe in these first few years as metallic drug-eluting stents.”

More Study, New Devices

Just how many of those trials and registries are currently enrolling patients and how many stents this entails have been slippery numbers to pin down. Even in the US, where the device has not been restricted to study scenarios, usage is rumored to be minimal. An Abbott spokesperson said she could not disclose individual product sales, but confirmed that the Absorb currently accounts for less than 1% of stent implantations in the United States.

Now, in the wake of this large and meticulous meta-analysis, it is equally unclear what outstanding questions, if any, can ethically be addressed using the current-generation device.

“I think at this point, we know the answers,” Wykrzykowska said. “I think we just need to go back to the drawing board and understand the data which we have already gathered. . . . When you put together the seven trials you have almost 6,000 patients, and there are even more patients treated in real clinical practice. So I think it’s enough to stop at this point and really analyze what we already have, what we already know, and follow the patients long-term, understand the mechanism, and indeed, invest time in developing something newer and better. At this point, I don’t think it makes sense to enroll more patients.”

Wykrzykowska predicts that within 5 or 6 years, investigators will have a “completely bioresorbable device that will meet all of our demands.” But this device will then need to jump through the hoops of first-in-human, CE Mark, and large, randomized controlled studies, to be followed by eventual commercialization. “That’s the nature of science: it all comes around in a big circle," she said. "But hopefully we can learn from the past.”

Jüni, speaking with TCTMD, reiterated that “the numbers needed to harm [in this meta-analysis] are just so low that it would not be advisable to use this device in clinical routines beyond very specific clinical trials.”

But pressed as to what these specific clinical trials should be addressing, Jüni came up empty-handed. The subgroup analyses in the current paper, while retrospective, probably provide sufficient answers to any niggling questions, in particular whether there are any subgroups who might benefit.

“It’s very hard to imagine a situation where the randomization of additional patients to this device would indeed contribute in a meaningful way beyond what we’ve seen in this meta-analysis,” Jüni said. “It might be much more important right now to focus entirely on next iterations of the scaffold, and careful evaluation first-in-animal studies and then carefully designed randomized trials.”

Note: Ali is an employee and Stone is a faculty member of the Cardiovascular Research Foundation, the publisher of TCTMD.