• Vascular Specialist Newspaper
• Volume 5 - 2009
• Number 9 online
• Number 8
• Number 7 online
• Number 6
• Number 5 Online
• Number 4 online
• Number 3
• Number 2
• Number 12 online
• Number 1
• Volume 4 - 2008
• Number 6
• Number 5
• Number 4
• Number 3
• Number 2
• Number 1
• Volume 3 - 2007
• Number 6
• Number 5
• Number 4
• Number 3
• Number 2
• Number 1
• Volume 2 - 2006
• Number 6
• Number 5
• Number 4
• Number 3
• Number 2
• Number 1
• Volume 1 - 2005
• Number 2
• Number 1

Home > Medical Professionals > Vascular Specialist newspaper > Vascular Specialist Volume 3 Number 1 > Vascular Specialist Volume 3 Number 1

Many clinicians might accept the results of the randomized European endovascular aneurysm repair (EVAR) trials at face value because they are level I evidence, but significant changes in management require a careful look at all of the data to see if they apply to clinical practice in the United States.

EVIDENCE SUGGESTS THAT EVAR SHOULD BE OFFERED TO OLDER PATIENTS AND/OR THOSE WITH SIGNIFICANT COMORBIDITIES AND LIMITED LONGEVITY.

The EVAR 1 and DREAM (Dutch Randomized Endovascular Aneurysm Management) trials randomized basically the same type of patients--those who had a large abdominal aortic aneurysm (AAA), had an anatomy suitable for EVAR, and were "fit" enough for open repair. The design and characteristics of patients in the DREAM trial were similar to those of the EVAR 1 trial, but the DREAM trial enrolled only 351 patients, compared with 1,082 in EVAR 1. Yet their findings and conclusions were more similar than different. There was no difference in the primary end point of all-cause mortality, but perioperative mortality was about 3% lower with EVAR than with open repair. At 2 years in DREAM and at 4 years in EVAR 1, AAA-related mortality was 3% lower with EVAR than with open repair. EVAR patients in DREAM reported a significantly higher quality of life than did openly repaired patients during the initial 6 months, but this difference lost significance after this initial period, beyond which it was similar to EVAR 1. EVAR entailed significantly higher complication and reintervention rates and higher overall costs than did open repair in both trials.

Even though the EVAR 1 trial, like DREAM, found no difference between EVAR and open repair in the primary end point of all-cause mortality, and showed that EVAR was worse or no better than open repair in four of five secondary end points (complication rate, reintervention rate, quality of life, and mean hospital costs), most of what we hear is about the 3% reduction in aneurysm-related mortality with EVAR that was maintained at 4 years in EVAR 1.

Following are some other aspects to consider before accepting the bottom line outcomes of these trials:

• Lower-risk patients with fewer comorbidities and with a longevity outlook that exceeds the 4-year cut-off point for analysis probably are not proportionately represented in the EVAR 1 trial's reported outcome data. Only 24% of patients had finalized outcomes at 4 years; 72% were still alive when the data were censored for the midterm report. Thus, the long-term follow-up data, due to be reported around 2010, may require a modification of the midterm conclusions.
• A total of 23 patients in EVAR 1 died before receiving their assigned treatment, probably because of a delay in treatment that averaged 57 days. For large AAAs with a mean diameter of 6.7 cm, such delays would be unlikely in current U.S. practice.
• Some other observations seem at odds with current U.S. practice. EVAR 1's open repair mortality of 4.7%, and the DREAM trial's 4.6%, seem higher than that of U.S. centers. Also the 54% of patients who were deemed to have unsuitable anatomy for EVAR appears high, since most U.S. centers report that only about 25%-33% of patients have unsuitable anatomy for EVAR. The difference in the latter may lie in the size of the AAA being treated. A recent Dallas study found that the rate of anatomic suitability for EVAR decreases fivefold for each centimeter increase in AAA diameter (Vasc. Endovascular Surg. 2005;39:381-91). The data from the EVAR 1 trial appear to dovetail with these findings. Further, the AAAs in EVAR 1 were all 5.5 cm in diameter or larger, whereas smaller AAAs are commonly treated by EVAR in the United States (as evident in the Food and Drug Administration device trials). Further, only 1.4% of those considered for enrollment in EVAR 1 were deemed "unfit" for open repair, a surprising contrast with U.S. practice where high risk is one of the commonest indications for choosing EVAR over open repair.
• After analyzing the results of the EVAR 1 and DREAM trials, I'm left with these questions:
• Is a 3% lower AAA-related mortality, but no overall mortality benefit, worth the higher complication and reintervention rates, costs, and lack of improvement in quality of life? In considering this, it should be noted that AAA-related mortality may be too "soft" of an end point to stand on its own because it assigns nearly all unwitnessed deaths to non-AAA-related causes. Many of the deaths in trials occur away from centers and few patients are autopsied nowadays. Thus, this end point helps to preserve the initial mortality advantage observed for EVAR in the perioperative period of the trials.
• Based on the results of these trials, should we choose EVAR over open repair for all "fit" patients with large AAAs and suitable anatomy, or for only those older patients with comorbidities and an expected longevity commensurate with the known durability of the stent graft?
• Are the mortality rates reported by the European trials generalizable to the United States or do they reflect some significant differences in AAA management in the major centers? One remembers the 5.8% mortality for open repair in the U.K. Small Aneurysm Trial being much higher than the 2.1% mortality in the Veterans Affairs' Aneurysm Detection and Management trial, so this is not the first time this question has arisen. This is a potentially important consideration because EVAR's advantage in the European trials is based on a high operative mortality for open repair, not on a low operative mortality for EVAR.

Similar concerns have arisen in regard to the EVAR 2 trial, which has drawn more criticism. This trial randomized 338 patients who were deemed "unfit" for open repair to either EVAR or no intervention. It found no differences in all-cause or AAA-related mortality between the two randomized groups. But proponents of EVAR have suggested that this trial's high crossover rate from no intervention to EVAR (27%, 47 of 172) and the long delays in those assigned to EVAR receiving this treatment may have biased the results against EVAR. For example, of the patients who were randomized to EVAR, 16 did not receive any intervention; 14 of those patients died before undergoing EVAR.

Because of such concerns, the EVAR 2 trialists reported performing per-protocol analyses and found no significant differences in either all-cause or AAA-related death between the two groups. If one also eliminates those patients who died awaiting EVAR and performs a post hoc analysis on the two remaining treatment groups, the all-cause mortality would appear to be significantly higher among patients who had no intervention (46%, 57 of 125 vs. 36%, 71 of 197). But, when one takes away from the denominator those 16 patients (4 randomized to EVAR and 12 of the crossover patients) who underwent open repair rather than EVAR, the difference between the EVAR and no treatment groups is only marginally significant (46%, 57 of 125 vs. 39%, 71 of 181). These post hoc analyses are therefore inconclusive in confirming a benefit for EVAR in high-risk patients that might have been obscured by trial irregularities.

The 166 patients who were assigned to receive EVAR had a perioperative mortality of 9%, which was much higher than the 2% (1 of 47) perioperative mortality among crossover patients. This begs the question: Does the high initial mortality for EVAR patients reflect delays in receiving EVAR or poor preprocedural patient preparation? And which EVAR mortality rate is closer to U.S. practice for high-risk patients?

An initial mortality of about 3% in high-risk patients may be more realistic for EVAR centers in the United States than the 9% reported by EVAR 2, if one considers subsequent reports from the United States for EVAR in high-risk patients; for example, 2.9% in the Lifeline Registry (J. Vasc. Surg. 2006;44:229-36), 3.1% in a large Veterans Affairs NISQIP study, and less than 3% in a high risk cohort at the Cleveland Clinic.

Finally, in a cost-effectiveness analysis of the European EVAR trial scenarios (Br. J. Surg. 2005;92:960-7), each quality-adjusted life-year for EVAR for patients in the EVAR 1 and DREAM trials cost about $200,000; that is not cost-effective. But a quality-adjusted life-year for EVAR for patients in the EVAR 2 trial cost about $15,000, which is well within the realm of what is considered cost-effective.

To summarize, my current view of the evidence suggests that EVAR should be offered to older patients and/or those with significant comorbidities and limited longevity, one which is within the limits of the reported durability of current stent grafts. These probably constitute a majority of patients whom we see in practice. The concern over device durability is moot in patients with limited longevity. However, because stent-graft durability remains an issue, I feel that open repair should be offered to "fit" patients with an extended longevity outlook. I still prefer EVAR over no intervention for high-risk patients, its originally intended use, but only after intensive in-hospital treatment of comorbidities. I cannot base this view on the results of the EVAR 2 trial but note that Dr. Janet Powell of Imperial College, London, and other EVAR 2 trialists have also concluded that the emphasis should be on improving fitness in such patients before deciding on treatment.