Long-term endograft data open to interpretation


Repair of abdominal aortic aneurysms (AAAs) can be undertaken through open surgery or endovascular repair (EVR). The first aortic stent graft was implanted as an investigational device in the US in 1993 and the FDA approved the first generation device in 1999. At the start of 2006 the number of devices approved had risen to five. There is a lot of optimism about aortic stenting, but what are the long-term results and what studies are used to support the facts? Dr Jon Matsumura, Associate Professor of Surgery, Northwestern University Feinberg School of Medicine, presented the available evidence to date.

There are various issues associated with EVR, noted Matsumura. These include: failure to reach the right location or to correctly deploy the stent graft; blood leaking between the graft and the vessel; dilation of the aortic neck; rupture after endo-repair; costly and frequent surveillance; and a need for reintervention. There are also problems that can occur with the stent grafts themselves: they can migrate after implantation, a limb can become occluded, it can permit flow through the material – ultrafiltration, and it can fracture or tear. Some of these problems have been recorded in papers as early as 1997, Matsumura added.

An overview of the industry-sponsored trials to date is, on the face of it, impressive reading. In the trials, the majority of utility measurements – at least those that were recorded – were significantly improved in the test subjects compared to controls, including procedure time, blood loss, intensive care stay, length of hospital stay, return to normal activity, and complete recovery at six months. Furthermore, he said, incidence of major adverse events at 30 days was on the whole lower.

Mentioning no trade names, Matsumura started with studies of a unibody polyester graft. These included 111 controls and 573 patients in the second set of bifurcated trials, including 319 in long-term follow-up. Of the 319, nine patients (2.8% using a non time-dependent calculation) were converted to surgery for a variety of reasons. In the follow-up cohort there were no device ruptures, although there was one in the original group of 573, and 20 have been reported worldwide to date. Furthermore, ten devices worldwide have type III endoleaks – from a defect in the graft. However, the longer the patients are followed for, the bigger the difference between them and the controls. While at 12 months the survival rate was 93.6% for patients and 94.4% for controls, this had dropped to 88.4% and 91.3% respectively at 24 months, and 69.5% versus 77.7% at 60 months, he said. “This trend is concerning, but it is a non-randomized study.”

For modular polyester grafts, 66 control patients had open surgery, with only one year’s follow up, while 1,192 patients with implants. The article showed there was a significant late conversion rate, noted Matsumura. This was for multifactorial reasons, including migration and material fatigue. This, he added, presents a serious risk of rupture. A retrospective analysis published in the Journal of Vascular Surgery (2000; 31: 960-970) reported: “Rupture appears to be related to poor device fixation.” Patient selection plays an important role within that: those whose aortas are severely tortuous with short angulated necks or whose infrarenal neck diameter is outside 16-26mm and less than 1.5cm long are unsuitable for stenting. Four-year primary endpoint analysis of this study once again used a flat-rate calculation. Results, published in a supplement to the Journal of Vascular Surgery (2001; 33: S135-145), showed that for 1,192 subjects, aneurysm rupture rate was 0.8%, death from aneurysm rupture was 0.3%, overall survival was 89.7%, and the late surgical conversion rate was 2.8%. “However, flat-rate calculations don’t capture what actually goes on,” observed Matsumura. The same data, analyzed using time-dependent Kaplan-Meier, show that the aneurysm rupture rate was 2.8%, aneurysm-related death was 3.4%, survival was 70%, and surgical conversion was 7.2%.

A pattern of late results that is more revealing is shown in the modular PTFE trials, said Matsumura. Radiographic corelab study showed that the proportion of subjects with AAA sac enlargement increased with each year of follow up to a marked proportion of 42% at five years. Looking at the cumulative graph of major adverse events, the difference between controls and test subjects persisted for the whole five years. While there was no aneurysm rupture at 60 months, 42% of subjects had aneurysm sac enlargement. This led the company to change the graft material in 2004.

The next trial discussed was for a modular polyester/suprarenal stent. The study enrolled 80 control patients, 200 standard-risk subjects and 100 high-risk subjects. Thirty-day mortality for the controls was 2.5%, compared to 0.5% of standard-risk patients and 2.0% of high-risk. One-year patency was very high for both test groups while one-year migration was 2.5% for standard and 2.8% for high-risk patients. In the trial, rate of material failure was low but not zero. Additional sutures were placed in the graft post-approval. The reintervention rate was 11-12% at one year; endoleak at 30 days was 17-18% for both test groups – comparable to other stent trial results – and fell to 4.8-5.0% at one year, highlighting the fact that lessons are being learned about how best to reintervene, he said.

Finally, Matsumura talked about the unibody PTFE stent design, approved by the FDA in November 2004. “It is interesting to note that there were a large number of small AAAs in the test group compared with the control group,” he commented. The mean size of the aneurysm in controls was 5.8cm compared to only 5.1cm in the test group, despite the fact that Cleveland Clinic data show that the risk of death is much higher in patients with aneurysms above 5.5cm. This study does not have any long-term results available.


Away from industry-sponsored trials, the randomized control trial is king. The EVAR 1 trial at 41 UK hospitals enrolled 1,082 elective AAA patients: 983 men, mean age 74 years and mean aneurysm size 6.5cm. Thirty-day mortality from endovascular repair was 1.7% compared to 4.7% for open surgery. Length of hospital stay was seven days for EVR compared to 12 for open surgery (Lancet 2004; 364: 843). At four years EVAR 1 showed that mortality for both groups was similar, at around 28%. Death related to aneurysms was only 4% for EVR compared to 7% for open surgery, but conversely, complications (not necessarily requiring reintervention) were 41% compared to 9% for surgery (Lancet 2005; 365: 2179).

The EVAR 2 trial covered 31 UK hospitals, and enrolled 338 patients deemed unfit for surgery at least 60 years old with AAAs at least 5.5cm. One hundred and sixty six patients had EVR while 172 had no intervention. At four years, mortality was similar for both groups at 64%, as was aneurysm-related mortality, thus implying that unfit patients get no benefit from EVR (Lancet 2005; 365: 2187).

Another, smaller, randomized control trial is the Dutch DREAM trial. This enrolled 345 patients with AAAs greater than 5cm. At 30 days, combined mortality and severe morbidity was 9.8% versus 4.7%, while mortality alone was 4.6% versus 1.2%. At two years, survival was similar for both groups and aneurysm-related mortality was 5.7% versus 2.1%, although this difference was non-significant (New England Journal of Medicine 2005; 352: 2398).

An interesting observation, said Matsumura, is that every device that has been tested and received FDA approved has since changed either its material or delivery system. Do these changes offer patients any enhancements? To determine that it is essential that trials assessing these new devices be carried out. Two such trials are ACE in France, which started in 2002 and covers 51 centers, 600 patients, and the US-based OVER trial, funded by the Department of Veterans Affairs, which will take eight years and has so far recruited more than 500 patients.

In conclusion, said Matsumura, “In anatomically suitable patients with large AAA, EVR is an alternative treatment because patients have a faster and more complete recovery, there are fewer complications, and there is a lower peri-operative mortality rate – although there is no real difference in overall survival.” However, EVR is more expensive, requires more reinterventions, and surveillance needs to be appropriately adjusted to take account of device-specific failure. Matsumura also added a corollary: “Be aware that short-term EVR data always look better than longer term data.”