SAFE-AAA Flags Ongoing EVAR Device Risk and Urges More Surveillance

Long-term monitoring of all aortic stent graft devices on the US market is long overdue, says Philip Goodney.

SAFE-AAA Flags Ongoing EVAR Device Risk and Urges More Surveillance

An endovascular stent graft for abdominal aortic aneurysm (AAA) that has had a troubled safety history does not meet noninferiority compared with other commercially available options, according to the SAFE-AAA analysis.

The study adds another chapter in what is becoming a long-running saga for patients treated with this class of device and raises pressing questions about the handling of safety signals and the need for surveillance for AAA patients in the years to come.

The US Food and Drug Administration first became aware of a rise in Type III endoleaks in grafted AAA patients through the voluntary Medical Device Reporting (MDR) system database. For a while, the agency believed the increase was associated with multiple devices, but in an updated safety letter in 2018 it was said to be specific to the only one with a unibody design, the AFX with Strata, which Endologix stopped producing in 2014. The manufacturer then changed the graft material, marketing the device as AFX with Duraply, and introduced the newer AFX2 graft.

While both AFX with Duraply and AFX2 initially seemed to be associated with fewer endoleaks, by the end of 2018, the entire AFX family of devices were under a Class I recall, the most serious type.

Now, SAFE-AAA, a large analysis of US Medicare patients, has found that at a mean follow-up of 3.3 years (maximum 8.4 years), the composite primary endpoint of aortic reintervention, rupture, and death occurred in 73.4% of patients treated with a unibody device versus 65% of those given a non-unibody device (P for noninferiority = 1.00).

The replication of the harm signal in the Medicare database supports what the FDA and its advisory panel had seen in smaller and voluntary reporting databases regarding the AFX platform and the risk of Type III endoleaks, said SAFE-AAA lead author Eric A. Secemsky, MD (Beth Israel Deaconess Medical Center, Boston, MA).

“The idea was how do we create an ecosystem for evaluating the long-term safety of these aortic stent grafts now that they've become the dominant approach for treating aortic aneurysm,” he added. “They were approved on shorter-term data, with most data sets going to about 5 years [during a time when] open AAA repair was still a primary approach.”

That’s a particularly concerning problem, he noted, for younger patients who may be candidates for endovascular aneurysm repair (EVAR) and who need longer-term data commensurate with their longer expected survival than older patients when deciding between it and surgery.


The study included 87,163 Medicare patients (mean age 77 years; 21% female; 3.7% Black) who underwent aortic stent grafting at 2,146 US hospitals from 2011 through 2017. Of these, approximately 14% received a unibody device. Compared with other EVAR devices on the market, the unibody stent grafts had a unique CPT code for identification in the Medicare claims database.

Time-varying analyses suggested that the association of the primary outcome and unibody devices was  significant at year 1 postimplantation, peaking at year 4. Compared with non-unibody devices, the unibody stent grafts were associated with greater risk of endograft extension (HR 1.31; 95% CI 1.01-1.69) and graft relining (HR 2.06; 95% CI 1.55-2.74), with no differences in late rupture, conversion to open repair, or all-cause death.

Additionally, in a prespecified analysis of approximately 2,700 patients that only included aortic stent grafts placed between February 22, 2016, and December 31, 2017—the time period when the AFX2 system was first commercialized— the cumulative incidence of the composite primary endpoint was 37.5% in unibody-treated patients versus 32.7% in those treated with non-unibody devices at a median of 2.6 years (HR 1.06; 95% CI 0.98-1.14).

“That we were seeing statistically significant associations specific to the newer unibody aortic sent graft, AFX2, more so than other market-available devices, suggest that maybe they have not fully resolved the problem,” Secemsky noted.

The FDA currently recommends at least yearly, lifelong follow-up for patients who have had EVAR with any AFX endovascular graft system to monitor for Type III endoleaks. The agency was convinced enough by the pre-publication SAFE-AAA results to consider them when making their decision to issue an updated safety communication in December 2022. In it, the agency said it had mandated that Endologix perform a postmarket study comparing real-world outcomes for patients implanted with AFX2 to those receiving other commercially available AAA endovascular grafts, with follow-up through 10 years.

Surveillance Options and Goals

The SAFE-AAA findings are in line with a study published in 2022 in the BMJ that compared long-term outcomes in Medicare patients who underwent EVAR with the AFX device or with one of three others: the Excluder (Gore), the Endurant (Medtronic), and the Zenith (Cook Medical). Overall, the AFX device, particularly the earlier iteration, failed more often than the others, with the suggestion that it was happening fairly consistently over time. In a subgroup of patients who received the newer AFX2, there were no differences in reintervention or late rupture compared with the other groups, but the follow-up was short at less than 3 years.

To TCTMD, that study’s lead author, Philip P. Goodney, MD, MS (Dartmouth-Hitchcock Medical Center, Lebanon, NH), said both the work of he and his co-authors and the SAFE-AAA study show why long-term surveillance of all aortic stent graft devices on the market is long overdue.

“It’s clear that stronger efforts are needed in terms of active surveillance for these devices after implantation to ascertain how they perform under real-world circumstances,” he said.

Secemsky cited the Society of Thoracic Surgeons/American College of Cardiology TVT Registry as one example of safety surveillance that includes all stakeholders and that may be a jumping off point for an aortic stent graft surveillance program. There is also the VISION Coordinated Registry Network, which consists of the Society for Vascular Surgery Vascular Quality Initiative (SVS VQI), the FDA, and other stakeholders. Claims-based data like those found in Medicare may complement those efforts, he added. 

Goodney and colleagues from the SVS VQI have proposed the Long-term EVAR Assessment and Follow-up (LEAF) system, which would be undertaken in conjunction with the FDA, industry, and professional societies representing the surgical, radiology, and cardiology communities.

LEAF would include details like why the graph was placed, relevant anatomical information, and other clinical and patient-specific variables, along with Medicare claims data.

“Many people use the AFX device for patients with occlusive disease, not aneurysmal disease. Those are different patients that perform differently over time in terms of their results,” he noted. “Clinical details matter, especially when we try to make device-specific conclusions. Those details should be present not just for devices, but for the patients in whom they're implanted . . . to aid in risk stratification and to provide a fair evaluation of devices in a prospective manner.”

  • The study was funded by the US Food and Drug Administration.
  • Secemsky reports consulting/speakers’ boards fees from Abbott, Bayer, BD, Boston Scientific, Cook, CSI, Endovascular Engineering, Janssen, Medtronic, Philips, VentureMed; and grants to his institution from BD, Boston Scientific, Cook, Laminate Medical, Medtronic, and Philips.
  • Goodney reports no relevant conflicts of interest.